Fabrication, Design, and Analytical Applications of Nanostructured Plasmonic Crystals

FABRICATION, DESIGN, AND ANALYTICAL APPLICATIONS OF NANOSTRUCTURED PLASMONIC CRYSTALS

SOMI KANG

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Materials Science and Engineering in the Graduate College of the University of Illinois at Urbana-Champaign, 2014

Urbana, Illinois

Doctoral Committee:

Professor Ralph G. Nuzzo, Chair Professor Paul V. Braun Professor John A. Rogers Professor Nancy R. Sottos

ABSTRACT

Surface plasmon resonances (SPRs) are coherent oscillations of electron density occurring at the interface of a metal and a dielectric, which generate an evanescent electric field that decays exponentially within ~100-200 nm from the surface of the metal. Because this enhanced electromagnetic field is highly sensitive to local optical property changes, SPRs have been exploited for real-time, fully label-free form of chemical/biological sensing and imaging or for field-enhanced applications of electronics and photovoltaics. Soft nanoimprint lithography provides inexpensive and versatile replication method to generate uniformly ordered and defined nanostructures over large areas. Plasmonic crystals consisting of squared arrays of nanoholes with high fidelity are fabricated using soft nanoimprint lithography, which exhibit the great potential for analytical applications. The work presented in this dissertation focused on the enhancement of analytical sensitivity of a new class of bimetallic plasmonic crystal and the development of plasmonic imaging technique for complex biomolecular system using nanostructured plasmonic crystal platform.

Bimetallic plasmonic crystals were demonstrated as a more sensitive substrate for quantitative bulk-refractive-index (Bulk IR) sensing and surface-enhanced Raman spectroscopy

(SERS) compared to mono-metallic plasmonic crystals with the same design rule. The best performances for each application of multispectral and SERS-based sensing were obtained by manipulating the composition of thin metal film, their spatial distribution, and the design rules of the plasmonic crystals. Finite-Difference Time-Domain (FDTD) simulations were used to verify the optical behavior of bimetallic plasmonic crystals and to understand the optimized device form factor.

A label-free optical imaging technique using plasmonic crystal was developed to quantitatively investigate the morphology and dynamic vital activities of cell. Polyelectrolyte layer-by-layer assemblies with well-defined thicknesses were used to calibrate the reflection contrast response as a function of thickness of biomolecular thin film on plasmonic crystal, which was theoretically verified through FDTD calculations. As a model system, Aplysia

California pedal neurons were cultured on plasmonic crystals and quantified in both dry and liquid conditions. The capability of this plasmonic imaging technique that investigates interaction between cell and substrate in real time was verified by cell detachment using trypsin treatment.

iii

To my family

ACKNOWLEDGMENTS

I believe that no significant achievements are accomplished entirely on one’s own. This has been the case in my own life as well. I would like to thank all the people who have helped and supported me to complete this dissertation.

I would like to begin by expressing my deepest gratitude and appreciation to my thesis advisor Prof. Ralph Nuzzo for his guidance, patience, and encouragement. His technical and editorial support has taught me innumerable lessons and insights on the workings of academic research. Ralph has granted me the freedom to design my thesis project and made me seek answers to scientific problems, which stretched my ability to look at the bigger picture.

I gratefully acknowledge Prof. Paul Braun, Prof. John Rogers, and Prof. Nancy Sottos for being in my thesis committee and for encouraging and guiding me during my graduate career. I am particularly grateful for the assistance given by Prof. Jonathan Sweedler whose expertise in neurobiology helped me achieve success in my cell imaging project.

I would also like to thank all the members of Nuzzo group - Sergio Sanchez, Audrey

Bowen, Matthew Smalls, Evan Erickson, Eric Brueckner, Enes Oruc, Lanfang Li, Jason

Goldman, Peixi Yuan, Michael Cason, Yuan Yao, Chris Corcoran, Junwen He, Mikayla Anderson,

Lu Xu, Chris Daily, Joselle McCraken, Marvin Malone, David Wetzel, Brittany Rauzan, Adina

Badea, Brian Lampert and Jinho Chang. I have been truly blessed to have worked with such an enthusiastic, talented, and invaluable group of people. I look forward to maintaining close contact with this network of colleagues.

I would also like to appreciate those who have collaborated with me during my graduate career. I am specifically thankful for the mentorship from Jimin Yao and An-Phong Le, who helped me start my own research. I am most grateful to Stanislav Rubakhin for sharing his

v knowledge of neuroscience. I would also like to extend my thanks to the staff members of Center for Microanalysis of Materials (CMM), Laser and Spectroscopy Facility (LSF), and

Micro/Nanofabrication Facility (Microfab), especially Julio Soares, Scott MacLaren, and Tao

Shang for their technical expertise and assistance.

Finally, I wish to thank my family and my friends for their love and support throughout my study. I deeply thank my parents and grandmother for encouraging me to choose what I desired and supporting me with their unconditional love. I also want to thank my lovely sister,

Soji Kang, who has been my best friend for my whole life. And special thanks to my 86 friends:

Iris Choi, Hajin Im, Jennifer Kim, Soyoung Kim, and Yoojin Kim. I thank all of you for your friendship and emotional support that enabled me to survive the harsh and lonely life in

Champaign.

TABLE OF CONTENTS

Chapter 1: Theory and Applications of Plasmonic Crystals ...... 1 1.1 Introduction ...... 1 1.2 Surface Plasmon Resonance Theory ...... 1 1.3 Soft Nanoimprint Lithography for Plasmonic Crystal Fabrication ...... 5 1.4 Finite-Difference Time-Domain Computational Modeling of Surface Plasmon Resonance ...... 8 1.5 Applications of Plasmonic Crystals ...... 11 1.5.1 Bulk Refractive Index Sensing Using Plasmonic Crystals ...... 12 1.5.2 Biological/Chemical Detecting Using Plasmonic Crystals ...... 13 1.5.3 Two-Dimensional Imaging Using Plasmonic Crystals ...... 15 1.5.4 Surface-Enhanced Raman Scattering Using Plasmonic Crystals ...... 16 1.5.5 Light Trapping for Photovoltaic Devices ...... 18 1.6 Overview of Dissertation ...... 19 1.7 References ...... 23 1.8 Figures...... 30

Chapter 2: Refractive Index Sensing and Surface-Enhanced Raman Spectroscopy Using Silver- Gold Bimetallic Plasmonic Crystals ...... 44 2.1 Abstract ...... 44 2.2 Introduction ...... 44 2.3 Experimental Materials and Methods ...... 47 2.3.1 Materials ...... 47 2.3.2 Plasmonic Crystal Fabrication via Soft Nanoimprint Lithography ...... 47 2.3.3 Bulk Refractive Index Sensing via Transmission-mode Spectroscopy ...... 48 2.3.4 Integrated Multispectral Response Calculation ...... 49 2.3.5 Surface-Enhanced Raman Spectroscopy (SERS) Measurements ...... 50 2.3.6 Finite-Difference Time-Domain (FDTD) Simulation of Plasmonic Nanostructures ...... 50 2.4 Results and Discussion ...... 51

vii

2.4.1 Structure and Properties of Bimetallic Plasmonic Crystals ...... 51 2.4.2 FDTD Modeling...... 54 2.4.3 Bulk Refractive Sensitivity of Bimetallic Plasmonic Crystals ...... 57 2.4.4 SERS Enhancement of Bimetallic Plasmonic Crystals ...... 60 2.5 Conclusions ...... 62 2.6 Acknowledgments...... 63 2.7 References ...... 64 2.8 Figures...... 68

Chapter 3: Quantitative Reflection Imaging of Fixed Aplysia Californica Pedal Ganglion Neurons on Nanostructured Plasmonic Crystals...... 80 3.1 Abstract ...... 80 3.2 Introduction ...... 81 3.3 Experimental Materials and Methods ...... 84 3.3.1 Materials ...... 84 3.3.2 Plasmonic Crystal Fabrication via Soft Nanoimprint Lithography ...... 84 3.3.3 Growth of Polyelectrolyte Layer-by-Layer Assemblies on Gold Films ...... 85 3.3.4 Ellipsometry of Polyelectrolyte Layer-by -Layer (LBL) Assemblies ...... 86 3.3.5 Cell Culture of Aplysia californica Pedal Ganglion Neurons ...... 86 3.3.6 Reflection Mode Plasmonic Imaging...... 87 3.3.7 Atomic Force Microscopy and Scanning Electron Microscopy of Cells on Plasmonic Crystals ...... 88 3.3.8 Finite-Difference Time-Domain (FDTD) Simulations of Plasmonic Nanostructures ...... 88 3.4 Results and Discussion ...... 89 3.4.1 Plasmonic Crystals and Biological System Model ...... 89 3.4.2 Reflection Imaging Contrast Calibration ...... 90 3.4.3 Reflection Imaging of Fixed Aplysia Pedal Ganglion Neurons on Plasmonic Crystals ...... 95 3.4.4 Atomic Force Microscopy (AFM) Height Profiles ...... 98 3.4.5 Increased Image Contrast through Wavelength Combination ...... 99

viii

3.4.6 Quantitative Thickness Estimates from Non-plasmonic Optical Effects ...... 102 3.4.7 Inferences Regarding Cell Structures Observed on Plasmonic Crystals ...... 105 3.5 Conclusions ...... 106 3.6 Acknowledgments...... 107 3.7 References ...... 108 3.8 Figures...... 112

Chapter 4: Quantitative Reflection Imaging for Morphology and Dynamics of Live Aplysia Californica Pedal Ganglion Neurons on Nanostructured Plasmonic Crystals ...... 129 4.1 Abstract ...... 129 4.2 Introduction ...... 130 4.3 Experimental Materials and Methods ...... 132 4.3.1 Materials ...... 132 4.3.2 Plasmonic Crystal Fabrication via Soft Nanoimprint Lithography ...... 133 4.3.3 Transmission Measurements of Plasmonic Crystals...... 134 4.3.4 Growth of Polyelectrolyte Layer-by-Layer Assemblies on Gold Films for Reflection Imaging Contrast Calibration ...... 134 4.3.5 Thickness and Refractive Index Measurements of Polyelectrolyte Layer-by- Layer (LBL) Assemblies in Artificial Sea Water (ASW) with Atomic Force Microscope (AFM) and Ellipsometry ...... 134 4.3.6 Reflection Imaging of Plasmonic Crystal Using a Laboratory Optical Microscope ...... 135 4.3.7 Reflection Contrast Calibration ...... 136 4.3.8 Cell Culture of Aplysia californica Pedal Ganglion Neurons ...... 137 4.3.9 Atomic Force Microscopy of Fixed Neuron Cells on Plasmonic Crystals ...... 138 4.3.10 Cell Dissociation from Plasmonic Crystal Surface Using Trypsinization ...... 138 4.3.11 Finite-Difference Time-Domain (FDTD) Simulations of Plasmonic Nanostructures ...... 138 4.4 Results and Discussion ...... 139 4.4.1 Plasmonic Crystals and Computational Models ...... 139 4.4.2 Index Corrected Mass Coverage ...... 140

4.4.3 Reflection Imaging Contrast Calibration ...... 143 4.4.4 Reflection Imaging of Live Aplysia Pedal Ganglion Neurons on Plasmonic Crystals ...... 146 4.4.5 Imaging the Effect of Trypsin on the Cell Activity ...... 150 4.5 Conclusions ...... 152 4.6 Acknowledgments...... 153 4.7 References ...... 155 4.8 Figures...... 158

Chapter 5: Future Research Directions for Quantitative Imaging Applications Using Nanostructured Plasmonic Crystals ...... 175 5.1 Sensitivity Improvement of Quantitative Plasmonic Imaging Technique for Live Cell ...... 175 5.2 Quantitative Imaging Technique Using Functionalized Plasmonic Crystals ...... 176 5.3 Characterizing Thickness and Refractive Index Using Two-Wavelength Analysis ...... 177 5.4 Final Remarks ...... 178 5.5 References ...... 179

CHAPTER 1

THEORY AND APPLICATIONS OF PLASMONIC CRYSTALS

1.1 Introduction

Surface plasmons are coherent electron oscillations that exist at the interface between metal and dielectric materials.[1, 2] The surface plasmons are excited by coupling with incoming light, which results in an evanescent electric field extending from the metal surface into the dielectric over a length-scale on the order of hundreds of nanometers.[3-6] This phenomenon is described as surface plasmon resonance (SPR). The evanescent electric field associated with SPR is highly sensitive to changes in near-interfacial optical profiles (refractive index or material thickness), which in turn provides a capability for label-free forms of SPR-based sensing and imaging.[7-11] In addition, the electromagnetic fields near a metal can be significantly enhanced via the coupling of incident electromagnetic radiation and surface plasmons.[12, 13] This attribute of SPR facilitates the use of field-enhanced applications including surface-enhanced

Raman scattering (SERS) and surface enhanced fluorescence.[14-17] Moreover, the evanescent field can localize and guide light on the surface of plasmonic structures. Due to this property, the study of SPR for light absorption enhancement in photovoltaic devices is also ongoing.[18, 19]

For example, preliminary results demonstrate enhancement in the photocurrent of thin-film solar cell using metallic plasmonic nanostructures.[20-22]

1.2 Surface Plasmon Resonance Theory

SPR is the collective oscillation of surface conduction electrons at a metal/dielectric interface by coupling with incident light that is evanescently confined in the perpendicular

1 direction over a few hundred nanometers.[3-6] There are two types of surface plasmon resonance: (i) surface plasmon polaritons (SPPs) and (ii) localized surface plasmonic resonance

(LSPRs).[15, 23] SPPs can propagate tens to hundreds of micrometers along the metal surface,[24] while LSPRs are non-propagating plasmon excitations that can be resonantly excited on metal nanostructures.[25] The difference between SPPs and LSPRs is illustrated in Figure 1.1.

The conventional SPR sensing methods employ the properties of SPPs that can be excited at the surface of flat metal film. By solving Maxwell’s equations, the dispersion relation for a

SPP ( kspp ) propagating along a semi-infinite planar metal/dielectric interface is given by

Equation 1.1 [26]:

 MD kspp  (1.1) c MD

where kspp is the wave vector of the SPPs,  is the angular frequency of the SPPs, c is the

speed of light,  M is the dielectric constant of metal, and  D is the dielectric constant of the surrounding dielectric environment. To meet the condition that surface plasmon waves are surface-confined, SPP mode should be transverse magnetic (TM) polarized waves and only exist at interfaces between materials with opposite signs on real portions of their dielectric

permittivities (i.e. between metal, Re M   0 , and dielectric,  D  0 ).[6] The momentum of a photon in air is generally less than that of the SPPs (illustrated in Figure 1.2). Thus, special momentum-matching techniques are required for coupling between incident photons and SPP modes.

One method to compensate for the momentum-mismatching is prism coupling using the

Kretschmann configuration (illustrated in Figure 1.3) to couple incoming light to a single surface plasmon resonance on a flat metallic thin film (~50 nm) based on the principle of total internal

2 reflection.[27] In the prism-based system, the changes in surface plasmon resonance due to the changes in the refractive index of the local environment can be identified by measurement of the intensity,[28, 29] resonant angle,[30] or resonance wavelength[31] of the reflected light wave.

One of the great advantages of a prism-based configuration is that the resulting data is easy to interpret because incoming light only couples with a single SPR mode in this system. However, prism-based SPR configurations consist of the prism, polarizers, lenses, and rotational stages, requiring precise alignment of all components. Because of this complex setup, prism-based momentum-matching techniques are difficult to miniaturize and to integrate with other devices.

In addition, improvements in lateral resolution are limited for prism-based SPR imaging systems due to the trade-off between the sensitivity and lateral resolution.[10] Long wavelength incident light (i.e. NIR) produces sharper SPR reflection for a given incident angle and a higher contrast

SPR image, while the propagation length of surface plasmons, which determines the lateral resolution of SPR imaging, increases as the incident wavelength increases.

The difficulties innate in using a prism-based system as a portable SPR imaging sensor can be resolved through the use of one-dimensional or two-dimensional metal grating structures, which diffract incoming light and thereby compensate for the momentum mismatch between the

SPP and the photon.[32, 33] The light diffraction facilitated by the grating provides the multiple additional momenta to incident photons, which accomplishes the coupling between incident light and SPP mode. This relation can be expressed using the following momentum conservation equation [34]:

kSPP k0 sin  iG x  jG y (1.2)

where kSPP is the wave vector of the SPP at a polar angle  , k0 is the wave vector of incident

photon, Gx and Gy are the Bragg vectors associated with the periodicity of the grating, and i and

3 j represent the order of scattering event. This momentum conservation relation shows that the grating-based system generates multiple photonic couplings to different SPP modes and that the coupling condition is adjustable by selecting the appropriate grating spacing and scattering orders. Therefore, the SPP behavior in grating-based systems requires a more complex interpretation as compared to the behavior observed for prism-based systems despite the improved portability in these systems.

One of the advantages imparted by metal grating-base system originates from the generation of LSPRs in which the surface plasmon is oscillated locally around sub-wavelength sized coupling objects.[23, 35-37] Although the shorter decay length of LSPRs lowers sensitivity to bulk refractive index changes associated with temperature fluctuations or solution concentrations,[15, 38] the LSPRs are more appropriately applied to SPR imaging with high lateral resolution due to their non-propagating nature. The lateral resolution of LSPRs can be improved to a single-nanoparticle limit,[39] while conventional prism-based systems limit the lateral resolution to tens of micrometers or more.[10, 40] Additionally, electric field intensity of

LSPRs is higher than that of SPPs, which allows for grating-based systems to be used as substrates for surface-enhanced spectroscopy techniques.

In addition to higher lateral resolution and stronger electric fields, a significant advantage of utilizing a grating-based system is the minimal hardware requirement. Since the grating-based system can couple photons to surface plasmons without complex optical alignment, these plasmonic nanostructures can be easily integrated into miniature or portable SPR devices at low cost. Moreover, the grating-based system exhibits greater flexibility for application-specific performance optimization because LSPRs produced using grating-based systems can be easily tuned by adjusting shape, size, composition, or assembly of metallic nanostructures.[12, 25, 41,

42] Due to the multi-faceted advantages of utilizing grating-based systems, square nano-hole arrays as platforms for SPR applications are central to a number of previous studies performed by our research group.[43-49] The details of the applications that we have previously reported using nano-hole plasmonic crystal arrays will be discussed in this chapter.

1.3 Soft Nanoimprint Lithography for Plasmonic Crystal Fabrication

Fabrication of metal nanostructures is an active area of study due to their interesting and extraordinary properties for diverse applications including photonics, optoelectronics,[50-53] and electronics,[54-56] as well as biochemical sensing and imaging.[23, 57-62] Nanostructures are typically formed using either a so-called ‘top-down’ or ‘bottom-up’ approach.[63, 64] Top-down techniques include conventional lithographic techniques such as photolithography,[65, 66] electron beam lithography,[67, 68] and ion beam lithography[69, 70] or unconventional methods such as block copolymer lithography,[71, 72] soft lithography,[73, 74] and scanning probe lithography.[75] Each of these techniques produces nanostructural motifs via the extraction of patterned nanostructures from bulk starting material. Bottom-up methods are used to construct nanostructures by controlling interactions of atoms, molecules, or more complex mesoscale objects.[76-80] This method facilitates the production of relatively smaller nanostructures with simple and inexpensive processing but is not appropriate for the complex design of nanostructures.[81] Therefore, the uniformly ordered and geometrically-precise structure required for plasmonic crystal preparation typically necessitates fabrication using a top- down methodology.

Photolithography is one of the typical, conventional methods used to mass-produce small features over large areas within the semiconductor industry.[82, 83] However, this method is not

5 ideal to make nano-scaled fine structures because a feature’s size is limited to the wavelength of light that is used.[84] Electron beam lithography (EBL) and focused ion beam (FIB) lithography can directly fabricate diverse nanostructures with precise control over size, shape, and spacing of metallic nanostructures. Those methods, however, are difficult to make infinite-area arrays of nano-patterns due to the slow processing speed and the high cost of instrument maintenance and operation.[64] Recent developments in unconventional types of lithography have provided new and effective techniques for patterning nanostructures on non-planar surfaces over large areas.

These techniques were performed at a low cost and were, therefore, appealing avenues towards synthesizing plasmonic devices.[23, 85]

A number of unconventional lithographic techniques were evaluated for application to plasmonic device fabrication, with each exhibiting specific advantages and limitations.

Nanosphere lithography uses a single layer of close-packed nanospheres as a deposition mask and is an inexpensive, versatile method to fabricate periodic arrays of plasmonic metal nanostructures over large areas of 10-100 μm2.[86-88] The periodic spherical voids created between neighboring nanospheres allow metal films to be selectively deposited or etched onto the substrate. Colloidal lithography is distinguished from nanosphere lithography by its random distribution of colloidal particles during deposition instead of the formation of a close-packed single layer.[23] The patterns of random particle distribution generated by this technique are capable of forming arrays of varying nanostructures such as nanoholes,[89] nanodisks,[90] or nanorings.[91] Unconventional polymeric fabrication techniques that have also been proposed include soft interference lithography and block copolymer lithography. Soft interference lithography can be exploited to create well-ordered nanostructures over large areas using polymeric phase mask for interfering light, and can be extended to prepare various nanostructural

6 forms such as infinite arrays, nanoparticles, or free-standing nanostructures through multiple exposure or etching steps.[36, 92] Block copolymer lithography can generate dense periodic patterns using self-assembling properties of block copolymer.[72, 93] Those unconventional lithography techniques are simple and high-throughput nanofabrication methods; however, the disadvantage of those methods is their limited pattern design. Scanning probe lithography allows for manipulation of various shapes of structures at the molecular level using scanning tunneling microscopy (STM), atomic force microscopy (AFM), or near-field scanning microscopy

(NSOM).[94, 95] Scanning probe lithography is not, however, appropriate for mass-production because it requires specific operation conditions and has limitations in processing speed.

Building upon previously proposed alternative lithographic methodologies, soft nanoimprint lithography was developed as a simple and versatile nanofabrication technique that allows for precisely defined nano-sized features generated using elastomeric stamps.[96-100]

Our research group has utilized this soft nanoimprint lithography to fabricate uniformly ordered nanohole arrays over large area at a low cost for a number of previously reported projects.[43,

46-49, 101] A schematic illustration of the soft nanoimprint process is presented in Figure 1.4a.

In brief, a composite hard-PDMS/soft-PDMS stamp is cast on a patterned photoresist master with arrays of nanohole relief structures and used to fabricate the plasmonic crystals. Composite

PDMS can improve mechanical properties and reusability of the elastomer stamp.[99] The

PDMS stamp is pressed into the molding material (e.g. UV-curable polyurethane, SU-8, and

Spin-On-Glass) on glass slide. After curing molding material using light or heat, the stamp is peeled away from the patterned polymer structure. Metal thin film is deposited onto this relief structure to complete the plasmonic crystal fabrication. In our previous reports, soft nanoimprint lithography was used to fabricate precise nanoholes arrays (optical and SEM images are shown

7 in Figure 1.4b and 1.4c) as the platforms for multiple applications such as chemical and biological sensing, SERS, and SPR imaging. These metallic nanohole arrays are termed plasmonic crystals because they contain highly uniform metal nanostructure periodicity over large areas. The optical properties of plasmonic crystals can be simply changed by modifying the lithographic process.[44, 49, 102] For example, the relief depth of a nanohole can be tuned by controlling the time and speed for spin-coating thin molding material film on glass slides. The resultant plasmonic crystals with different relief depths show different transmission properties.[49] In addition to changes in size or lattice parameters of nanoholes,[102] the change in distribution of a metal resonant film by adjusting metal deposition processes[48] is another simple way to manipulate properties of plasmonic crystals. More details on fabrication and optical property changes of nanoimprinted plasmonic crystals are herein discussed.

1.4 Finite-Difference Time-Domain Computational Modeling of Surface Plasmon

Resonance

Computational modeling of grating nanostructures is a powerful tool with which to elucidate the underlying physics and to predict the optical responses of devices with arbitrary design rules without the expense of physical fabrication. The optical behavior of plasmonic crystals is not easy to predict intuitionally because periodic grating structure generates multiple plasmonic features that include Wood’s anomalies, Bloch wave SPPs, and Fano resonance, as well as LSPRs around nanostructural features.[103] To theoretically characterize optical properties of plasmonic crystals, we have extensively simulated these structures using the finite- difference time-domain (FDTD) method.

The FDTD method was introduced by Yee in 1966[104] and has been developed as a versatile and accurate electromagnetic modeling tool to solve Maxwell’s equations.[105] To implement FDTD calculations, a computational model is first divided into many small cubes

(each of which is termed as a Yee’s cell), following which the electric and magnetic fields of each unit cell are determined using Maxwell’s differential equations. Although the grid spatial discretization can degrade the accuracy of computational results for curved structures, the time- domain solutions of FDTD computations can provide animated displays of the electromagnetic field distribution throughout the modeled system.[106, 107] This time-domain method of FDTD simulation is also able to calculate response of the model over a range of wavelengths from a single simulation and treat impulse or nonlinear behavior naturally.[105]

To compute optical features of infinite square arrays of nanostructures under electromagnetic radiation, the unit cell geometry is defined as a single nanostructure with appropriate boundary conditions. The unit cell is subdivided into a number of grid points (Yee’s cell) with an arbitrary spacing along the x , y , and z axes, and the interaction between nanostructure and electromagnetic fields is calculated at each grid point during a predetermined length of time. Defining appropriate optical parameters of modeled materials (permittivity or refractive index) is significantly important for more accurate results of FDTD modeling. For example, the Drude plus multi-pole Lorentzian model was widely employed to fit complex frequency-dependent permittivity of gold and silver which are the two most widely used materials for surface plasmon resonant layers.[108]

FDTD simulations can be exploited to theoretically understand the spectral behavior of plasmonic crystals and to visually demonstrate electric field distributions around nanostructures.

Figure 1.5 demonstrates the alteration of SPR features depending on minute structural

9 differences of plasmonic crystals using two nanohole arrays with different metal distributions that correspond to the quasi-3D and the full-3D plasmonic crystals. Figure 1.5a and 1.5b exhibit good correspondence between experimentally measured and computationally determined normal incident transmission spectra for quasi-3D nanohole and full-3D nanohole arrays, respectively.

Small discrepancies probably originate from a combination of effects, such as geometric limitations in the computational model, approximation of the wavelength-dependent refractive indices of the metal and dielectric, and nanoscale defects or roughness in the actual metal layer.

Electric field distributions associated with the plasmonic features labeled as B and C in transmission spectra of the quasi-3D and full-3D plasmonic crystals are illustrated in Figure 1.5c and 1.5d, which were computed to verify types of SPR mode of each plasmonic feature in spectra. The largest feature in the quasi-3D transmission spectrum (label C in Figure 1.5c) is assigned to BW-SPP or Wood’s anomaly excitations, while that in the full-3D transmission spectrum (label C in Figure 1.5d) is the result of LSPR excitations confined around the top rim of the nanohole. Also, we can verify that the second largest features in both transmission spectra of quasi-3D and full-3D plasmonic crystals are generated by different SPR modes, in Figure 1.5c, feature B shows electric field distribution of an LSPR mode, and in Figure 1.5d, feature B corresponds with a combination of BW-SPP and Wood’s anomaly.

FDTD calculations also can be used to optimize performance of plasmonic devices by evaluating the design rules of plasmonic crystals such as the relief depth, periodicity, nanohole diameter, and the distribution of metal resonant film.[109] Figure 1.6 demonstrates that the optimized system structure of squared nanohole arrays for refractive index sensing can be predicted using FDTD modeling. The figure-of-merit (FOM) for refractive index sensing was

determined by integration of normalized transmission differences ( TTTnorm  solution  baseline ) over

10 the collected wavelength range. Figure 1.6a shows the normalized transmission difference of calculated normal transmission spectra for a plasmonic crystal of 350 nm relief depth in solutions with different concentrations of polyethylene glycol (PEG). Despite the observation that magnitudes of experimentally determined FOMs are smaller than those calculated due to various imperfections on metal film, the changes to calculated FOMs as a function of periodicity shows similar gross trends with experimentally obtained FOMs (shown in Figure 1.6b). Similarly, the changes in calculated FOMs for plasmonic crystals with increasing gold film thickness (shown in

Figure 1.6c) can be used to estimate approximately the influence of gold film thickness on the experimentally determined FOM values of plasmonic crystals (in Figure 1.6d). Although it is difficult to quantitatively forecast the value of FOMs due to the experimental limitations, FDTD calculations are a robust tool to qualitatively predict the optical behavior of plasmonic devices and to optimize the structures of plasmonic systems in order to achieve their best performance.

1.5 Applications of Plasmonic Crystals

The periodic nanostructures in metal films have been utilized as a platform for a variety of SPR applications and studied extensively to facilitate understanding of the underlying physics of surface plasmon resonance. These efforts have been directed particularly toward the unique optical properties of periodic nanostructures such as extraordinary optical transmission and high sensitivity to optical changes in the local environments.[32, 110-112] Our research group in particular has focused on implementing square arrays of nanoholes fabricated using soft nanoimprint lithographic techniques to develop a diverse array of SPR applications including chemical/biological sensing and imaging, field-enhanced spectroscopy, and plasmonic solar cells.

1.5.1 Bulk Refractive Index Sensing Using Plasmonic Crystals

The surface plasmon resonance is extraordinarily sensitive to changes on refractive index of surrounding environment. Due to this nature, plasmonic crystals have been used as a bulk refractive index sensor. The refractive index sensitivities of SPR sensors have been commonly determined by measuring changes in position or intensity of single resonance peak. Although single resonance peak analysis method allows for simple interpretation of optical data, it is not suitable to evaluate the sensitivity of a plasmonic crystal for which optical responses are dictated by complex plasmonic behavior associated with the LSPRs, BW-SPPs, and Wood’s anomaly responses.[103] Therefore, the multispectral sensitivity analysis was introduced to determine the bulk refractive index sensitivity of plasmonic crystals.[46, 113-115]

Figure 1.7a displays the changes in transmission spectrum of a quasi-3D plasmonic crystal as the concentration of polyethylene glycol (PEG) solutions increase.[47] The surrounding refractive index was controlled as PEG solutions of varying concentration were injected onto the surface of a flow-cell mounted plasmonic crystal (shown in Figure 1.8). To consider both the peak shifts and the intensity changes over the collected wavelength range, the absolute value of differences between transmission spectra collected in each PEG solution and the spectrum recorded in pure water was integrated using the following equation:

R| (% T ( ) | d (1.3)

This integral response, R, has unit of %T nm and changes linearly as a function of the refractive index of the PEG solution (demonstrated in Figure 1.7b). This multispectral analysis method is not only more suitable to reflect various SPR modes, but also more precise for detecting changes in bulk refractive index as compared to common SPR-based measurements.

For example, signal-to-noise ratio of the integrated multispectral response of a quasi-3D

12 plasmonic crystal was improved by a factor of 3-10 times over that of a single-wavelength response of the same optical system.[47]

1.5.2 Biological/Chemical Detecting Using Plasmonic Crystals

In addition to bulk refractive index sensing, the multispectral sensitivity of evanescent electric fields offers interesting opportunities for identifying specific analytes of interest or monitoring dynamics of analytes. SPR-based sensing methods have been actively exploited to employ their unique ability for label-free biological/chemical detection under extremely dilute conditions in real time. For example, plasmonic crystals have been used as substrates for measuring the dynamic adsorption of biomolecules and chemomechanical forces.[43, 47, 116]

The application of full-3D plasmonic crystals to quantitative analytical bioassays was demonstrated using the binding of goat IgG in solution to immobilized antigoat IgG on the surface of a plasmonic crystal.[47] To immobilize and orient antigoat IgG, protein G was covalently coupled to the surface of the plasmonic crystal on which a carboxyl-terminated thiol self-assembled monolayer was applied. Figure 1.9a shows the normalized integrated transmission changes as a function of the concentration of goat IgG solution injected into a flow cell-mounted plasmonic crystal and verifies the substantive agreement between surface coverage of IgG on the plasmonic crystal (corresponding with normalized integrated response) and the

Langmuir isotherm model. The surface confined affinity constant is estimated as 7×107 M-1 by fitting data to a Langmuir adsorption model, which is consistent with previously reported results for IgG/anti-IgG systems.[117, 118]

Figure 1.9b exhibits the specific binding event of a biotin-avidin on quasi-3D plasmonic crystals.[43] Injection of biotinylated bovine serum albumin (bBSA) into a flow-cell mounted plasmonic crystal triggered an increase and subsequent plateau in the integrated response

(marked as (i) in Figure 1.9b) due to the adsorption of a bBSA monolayer onto the surface of plasmonic crystals. This bBSA monolayer was prevented from further nonspecific adsorption

(confirmed using phosphate buffered saline (PBS) and unmodified bovine serum albumin (BSA) solution, marked as (ii) in Figure 1.9b) on the sensing substrate, but was observed to specifically bind to avidin. This specific binding of surface-bound bBSA and avidin was observed from the large increase in integrated response after injecting avidin (marked as (iii) in Figure 1.9b).

Response of final binding of terminating layer of bBSA and surface-immobilized avidin was smaller than initial binding of immobilized bBSA and free avidin, consistent with layer- dependent mass coverage previously reported in similar assays.[119, 120]

The pH changes in the surrounding solution are extremely difficult to measure using SPR responses because these changes generate only small alterations in bulk refractive index (less than 1×10-4). Plasmonic crystals, however, can be used as pH sensors by coupling with a surface- bound pH sensitive hydrogel.[116] The pH gradients in the hydrating solution influence the force of osmotic pressure exerted on the hydrogel, and the resultant mechanical change in the hydrogel matrix is optically monitored by the plasmonic crystal. The top panel of Figure 1.9c shows the reversible changes in transmission of the hydrogel-modified nanohole arrays, after which analyte solutions was cycled between pH 7.86 and 1.44. A correlation between the magnitude of integrated transmission changes and pH changes is demonstrated in bottom panel of Figure 1.9c using three different pH change cycles; pH 7.86 to 1.44 (blue), 6.42 to 5.13 (red), and 5.76 to

5.66 (black). Remarkably, this hydrogel-modified plasmonic pH sensor is highly sensitive to pH changes in solution, even enabling recognition of changes down to 0.1 pH units without noise

(presented in magnified inset of Figure 1.9c).

1.5.3 Two-Dimensional Imaging Using Plasmonic Crystals

The grating-based SPR system capable of resolving surface adsorbates with higher spatial resolution[121, 122] provides potential for utilities in two dimensional chemical/biological imaging as well as spectroscopic sensing. In particular, the high structural uniformity over large areas allows nanoimprint plasmonic crystals to function as stable platforms for thin film imaging sensors under near-infrared or visible wavelengths.[46, 113, 123] An SPR imaging sensor built using plasmonic crystals and simple optical instruments such as a silicon charge-coupled device

(CCD) camera-equipped optical microscope is more applicable for real-time detection of biological and chemical agents. Ultimately, this imaging method is easier to design, is relatively inexpensive, has a more compact optical setup, and outputs visual data in a more intuitive than spectroscopic sensing.

A Quantitative thin film imaging technique using a nanoimprinted plasmonic crystal was also demonstrated using three different proteins (fibrinogen, -globulins, and myoglobin) that nonspecifically adsorbed to the surface of full-3D nanohole arrays (illustrated in Figure

1.10).[47] The patterned protein lines on the plasmonic crystal were imaged under white light illumination and were readily discerned from the background due to the different SPR responses in each region of discrete refractive indices. Moreover, this plasmonic crystal-dependent quantitative imaging method can be used as a molecular ruler. Figure 1.10b and Figure 1.10c show the thickness of three protein lines calibrated based on a contrast change of a 2 nm reference thin film (a change in image intensity of ~0.04 (0.01)), which is in good agreement with the ellipsometrically determined thickness of the protein on a planar gold film.[124]

Figure 1.11 validates the excellent sensitivity of the nanohole plasmonic imaging sensor, which facilitate discernment even of mass differences between two-carbon atoms.[46] The

15 patterns of 1-octadecanethiol (ODT) monolayers formed by a micro-contact printing method on a full-3D plasmonic crystal were clearly differentiated from the uncovered area due to refractive index differences between two regions (in Figure 1.11a). By immersing ODT-patterned plasmonic crystals into an ethanol solution of 1-hexadecanethiol (HDT), image contrast decreased as differences in refractive indices between the ODT pattern and the rest of HDT modified regions decreased, but ODT patterns were still distinctly distinguishable (in Figure

1.11b). The magnitude of decrease in image contrast at the endpoint of the backfilling experiment (at 300 s), which is about 10% of its initial value, correlates well with the 11% difference in the chain length of ODT and HDT. The plot in Figure 1.11c shows that the image contrast decreased as a function of backfilling time with HDT, which corresponded with a first- order Langmuir adsorption isotherm. Therefore, SPR imaging using plasmonic crystals was confirmed as a powerful method to quantitatively visualize and analyze ultra-thin films in real time without expensive optical systems.

1.5.4 Surface-Enhanced Raman Scattering Using Plasmonic Crystals

Raman scatting is defined as inelastic scattering of photons from a molecule in which the frequency of the photon is shifted by the energies of molecular vibrations.[125] Raman spectroscopy can be utilized for molecular characterizations because vibrational information is specific to the chemical bonds and symmetry of molecules. The main limitation of Raman spectroscopy is, however, weak intensity of signal, making it challenging to separate Raman scattering from the intense elastic scattering in certain experimental circumstances.

Enhanced electric fields that are associated with LSPRs and generated via plasmonic nanostructures can improve intensity of Raman signals by several orders of magnitude (as much as 10 to 11 orders of magnitude).[58, 126] Spectroscopic analysis using a Raman signal

16 amplified by surface plasmon resonance is termed surface-enhanced Raman spectroscopy

(SERS). The Raman signal produced by an early SERS system using roughed metal film or nanoparticles was vastly irreproducible owing to its ill-defined structures.[127] Therefore, plasmonic crystals which consist of uniformly ordered nanostructures have been exploited as more promising substrates for SERS sensing and imaging with reproducibly exceptional sensitivity.[102]

Figure 1.12a demonstrates the enhanced Raman intensities of benzenethiol using quasi-

3D plasmonic crystals with varying diameters and nanohole periodicities. According to the approximate requirement for SERS enhancement, the maximum enhancement is expected when the plasmonic crystal creates LSPR at wavelengths halfway between laser excitation (785 nm) and the Raman scattered wavelength of benzenethiol (857 nm corresponding to a Raman shift of

1073 cm-1), corresponding to 821 nm.[128, 129] The transmission intensities at 821 nm and the

SERS signal at 1073 cm-1 of sixteen different nanohole arrays with different periodicities

(presented in Figure 1.12b) are in qualitatively good agreement, which verifies correlation between LSPR strength and SERS enhancement.

Due to the long-term stability in SERS responses (over ~150 days in air) and high structural uniformity over large areas, plasmonic crystals can be employed as an effective substrate for SERS imaging. The top image in Figure 1.12c shows nanohole arrays patterned within the letters of “UIUC” on SU8 substrate. The map of Raman intensity at 1073 cm-1 from this SU8 substrate after soaking in benzenethiol (in the bottom panel of Figure 1.12c) demonstrates that high Raman signal is detected only in the areas where nanohole arrays are present, despite the presence of an adsorbed benzenethiol monolayer across the entire surface.

1.5.5 Light Trapping for Photovoltaic Devices

Thin film silicon (Si) solar cells have attracted attention as a next-generation energy source due to the economic efficiency and manufacturing feasibility for light-weight and flexible devices. Si is the dominant commercial material in the semiconductor and photovoltaic industry because of its abundance and non-toxicity as well as its exceptional optical and electronic properties.[130-132] However, Si with the indirect band-gap is a relatively weak light absorber, in particular over the 600-1100 nm spectral range. This limitation requires optically thick dimensions of conventional Si solar cells (~180-300 μm), which impedes the ability of Si solar cells to be widely used in affordable, practical applications. For lower system costs, alternative thin-film solar cells with light trapping structures have been developed.[133-135] The standard method for increasing light absorption in a solar cell is to employ pyramidal surface textures with a double-layered anti-reflection coating.[136-138] Although this structure can effectively reduce light reflection from the front surface and increase the optical path length by scattering light at multiple angles, micrometer-sized textured surface (from 5 to 10 μm) is not appropriate for thin film solar cell design. Therefore, light trapping schemes on the nano-scale are needed for the future of affordable photovoltaic technology.

Recently, plasmonic solar cells have been proposed to improve light absorption using metallic nanostructures that support surface plasmons.[21, 139-141] Plasmonic structures for concentrating and scattering light into the thin film solar cells can be engineered in three ways

(illustrated in Figure 1.13).[142] First, metal nanoparticles at the surface of the solar cell allow light to be scattered and folded into the solar cell. In addition, embedded plasmonic nanostructures in the solar cell can act as sub-wavelength antennas, thereby increasing the effective absorption cross section. Lastly, light can be trapped and guided by the excitation of

SPPs at the interface between semiconductor and the backside corrugated metallic film. These light-trapping techniques using plasmonic structures have great potential for development of thin film solar cell without loss of light absorption.

Figure 1.14 illustrates a demonstration of enhanced short circuit current densities of ultrathin hydrogenated amorphous Si (a-Si:H) solar cells combined with nanoimprinted plasmonic back-contacts at which SPPs are confined. The design of plasmonic solar cell with

Ag/ZnO:Al back-contact is illustrated in Figure 1.14a and a photograph of a finished solar cell is shown in Figure 1.14b. An SEM image of a Ag-coated silica sol-gel pattern (shown in Figure

1.14c) attests to the uniformity of nanoparticles with 290 nm diameters and 500 nm pitch.

Devices can be characterized by quantum efficiency (QE) measurements to evaluate the enhanced performance of plasmonic solar cell compared with similar solar cells on flat or randomly textured substrates. Current density/voltage measurements (in Figure 1.14d) demonstrate that the best cell having nanostructures with 250 nm diameter and 500 nm pitch increases in Jsc by 46% over that of the flat cell and has a high efficiency of 6.6%, which is similar to the maximum efficiency found for a thick cell. The effects of plasmonic structures on solar cell efficiency can be further verified using FDTD computations. The overall spectral correspondence between the EQE measurements (Figure 1.14e) and the electromagnetic simulation using FDTD (Figure 1.14f) strongly suggests that the enhanced short circuit current of plasmonic solar cell is due to increased light absorption by plasmonic structures.

1.6 Overview of Dissertation

This dissertation describes analytical applications for nanostructured plasmonic crystals consisting of square arrays of nanoholes fabricated using soft nanoimprint lithography. Soft

19 nanoimprint lithography provides a simple and powerful method to fabricate plasmonic nanostructures over large area with high spatial fidelity. The nanoimprinted plasmonic structures are capable of coupling between incident light and surface plasmons, which generate multiple plasmonic modes including propagating surface plasmon polaritons (SPPs) and localized surface plasmon resonances (LSPRs). The evanescent electric field originating from surface plasmon resonance is highly sensitive to small optical changes near the surface of metallic nanostructures.

Therefore, plasmonic crystals offer promising opportunities for spectroscopic chemical/biological sensing and SPR-based imaging applications.

Chapter 2 describes the development of an Ag/Au bimetallic plasmonic crystal as an optical system with superior sensitivity for quantitative bulk refractive index sensing and surface-enhanced Raman spectroscopy (SERS) to mono-metallic plasmonic crystals. Bimetallic plasmonic crystals exhibit outstanding performance that stems from distinctive plasmonic features and strong localized electric fields produced by both the Ag and Au metallic layers, as well as chemical stability imparted by the Au. To achieve the greatest improvement in refractive index sensitivity and Raman signal, the thickness ratio for each metal and the geometric parameters of the plasmonic crystal nanostructure can be optimized. Finite-Difference Time-

Domain (FDTD) simulations are then used to theoretically verify the underlying physics of surface plasmon resonances associated with multispectral refractive index and SERS-based chemical sensing. The results from this work demonstrate a robust and low-cost platform for chemical/spectroscopic sensing.

Chapter 3 describes the development of a reflection imaging technique capable of quantitatively determining the thickness of bio-molecular structures on the surface of a nanostructured plasmonic crystal in dry conditions. Polyelectrolyte films grown with a layer-by-

20 layer assembly method were imaged through bandpass filters and used to calibrate reflection contrast changes as a function of thickness. To compensate for refractive index differences of polyelectrolyte films and biomolecular systems, ellipsometric modeling and FDTD computation were performed to derive the refractive index-corrected layer thicknesses. The reflection contrast responses originating from surface plasmon resonances were applied for quantitative analysis of

Aplysia californica pedal neurons cultured and fixed directly on the plasmonic crystal surface, which demonstrates the potential application of plasmonic crystals as substrates for label-free forms of quantitative imaging techniques. Combining reflection images acquired using different wavelength ranges is suggested as a method with which to increase image contrast and sensitivity to topological differences. The limitation of plasmonic imaging techniques to a narrow measurable thickness range (less than 100 nm) was complemented with an interferometric analysis method. This simple reflection imaging technique therefore shows potential as a quantitative method for analyzing surface thicknesses at nanometer scales over large areas in real-time.

Chapter 4 describes the utilization of a reflection imaging technique for quantitative imaging of live cell morphology and cell-substrate interactions within biocompatible environments. In the same manner as described in Chapter 3, polyelectrolyte thin films are used as a reference model system and FDTD calculations are utilized to verify the reflection contrast changes caused by increasing the thickness of index-corrected materials for live biological systems. Studies conducted on complex biological systems such as the live Aplysia californica pedal ganglion neurons that were immersed in artificial sea water (ASW) served to illustrate the substantive potential for using plasmonic reflection imaging techniques as multi-functional analytical tools for studying dynamic living systems and organisms. In this study, the

21 morphology of peripheral structures (< ~80 nm) of live Aplysia pedal neurons was able to be quantitatively analyzed. Furthermore, a complex process such as cell detachment induced by trypsinization was visualized directly and interpreted. These results validate that plasmonic reflection imaging methodology can serve as a promising real-time analysis method for investigating quantitatively the morphology and dynamics of ultrathin cellular systems in complex environments.

1.7 References

1. Stern, E.A. and R.A. Ferrell, Surface Plasma Oscillations of a Degenerate Electron Gas. Physical Review, 1960. 120(1): p. 130-136. 2. Polman, A. and H.A. Atwater, Plasmonics: optics at the nanoscale. Materials Today, 2005. 8(1): p. 56-56. 3. Barnes, W.L., A. Dereux, and T.W. Ebbesen, Surface plasmon subwavelength optics. Nature, 2003. 424(6950): p. 824-830. 4. Knoll, W., Interfaces and thin films as seen by bound electromagnetic waves. Annu Rev Phys Chem, 1998. 49: p. 569-638. 5. Ozbay, E., Plasmonics: Merging photonics and electronics at nanoscale dimensions. Science, 2006. 311(5758): p. 189-193. 6. Maier, S.A., Plasmonics : fundamentals and applications2007, New York: Springer. xxiv, 223 p. 7. Liedberg, B., C. Nylander, and I. Lundstrom, Surface-Plasmon Resonance for Gas- Detection and Biosensing. Sensors and Actuators, 1983. 4(2): p. 299-304. 8. Homola, J., S.S. Yee, and G. Gauglitz, Surface plasmon resonance sensors: review. Sensors and Actuators B-Chemical, 1999. 54(1-2): p. 3-15. 9. Zhao, J., et al., Localized surface plasmon resonance biosensors. Nanomedicine, 2006. 1(2): p. 219-228. 10. Brockman, J.M., B.P. Nelson, and R.M. Corn, Surface plasmon resonance imaging measurements of ultrathin organic films. Annu Rev Phys Chem, 2000. 51: p. 41-63. 11. Steiner, G., Surface plasmon resonance imaging. Analytical and Bioanalytical Chemistry, 2004. 379(3): p. 328-331. 12. Kelly, K.L., et al., The optical properties of metal nanoparticles: The influence of size, shape, and dielectric environment. Journal of Physical Chemistry B, 2003. 107(3): p. 668-677. 13. Jain, P.K., et al., Review of some interesting surface plasmon resonance-enhanced properties of noble metal nanoparticles and their applications to biosystems. Plasmonics, 2007. 2(3): p. 107-118. 14. Campion, A. and P. Kambhampati, Surface-enhanced Raman scattering. Chemical Society Reviews, 1998. 27(4): p. 241-250. 15. Willets, K.A. and R.P. Van Duyne, Localized surface plasmon resonance spectroscopy and sensing. Annu Rev Phys Chem, 2007. 58: p. 267-97. 16. Neumann, T., et al., Surface-plasmon fluorescence spectroscopy. Advanced Functional Materials, 2002. 12(9): p. 575-586. 17. Roy, S., et al., Surface plasmon resonance/surface plasmon enhanced fluorescence: An optical technique for the detection of multicomponent macromolecular adsorption at the solid/liquid interface. Langmuir, 2002. 18(16): p. 6319-6323. 18. Stuart, H.R. and D.G. Hall, Island size effects in nanoparticle-enhanced photodetectors. Applied Physics Letters, 1998. 73(26): p. 3815-3817. 19. Schaadt, D.M., B. Feng, and E.T. Yu, Enhanced semiconductor optical absorption via surface plasmon excitation in metal nanoparticles. Applied Physics Letters, 2005. 86(6). 20. Pillai, S., et al., Surface plasmon enhanced silicon solar cells. Journal of Applied Physics, 2007. 101(9). 21. Catchpole, K.R. and A. Polman, Plasmonic solar cells. Opt Express, 2008. 16(26): p.

21793-21800. 22. Akimov, Y.A., K. Ostrikov, and E.P. Li, Surface Plasmon Enhancement of Optical Absorption in Thin-Film Silicon Solar Cells. Plasmonics, 2009. 4(2): p. 107-113. 23. Stewart, M.E., et al., Nanostructured plasmonic sensors. Chem Rev, 2008. 108(2): p. 494-521. 24. Barnes, W.L., Surface plasmon-polariton length scales: a route to sub-wavelength optics. Journal of Optics a-Pure and Applied Optics, 2006. 8(4): p. S87-S93. 25. Hutter, E. and J.H. Fendler, Exploitation of localized surface plasmon resonance. Advanced Materials, 2004. 16(19): p. 1685-1706. 26. Raether, H., Surface-Plasmons on Smooth and Rough Surfaces and on Gratings. Springer Tracts in Modern Physics, 1988. 111: p. 1-133. 27. Kretschm.E and H. Raether, Radiative Decay of Non Radiative Surface Plasmons Excited by Light. Zeitschrift Fur Naturforschung Part a-Astrophysik Physik Und Physikalische Chemie, 1968. A 23(12): p. 2135-&. 28. Nylander, C., B. Liedberg, and T. Lind, Gas-Detection by Means of Surface-Plasmon Resonance. Sensors and Actuators, 1982. 3(1): p. 79-88. 29. Manuel, M., et al., Determination of Probable Alcohol Yield in Musts by Means of an Spr Optical Sensor. Sensors and Actuators B-Chemical, 1993. 11(1-3): p. 455-459. 30. Liedberg, B., I. Lundstrom, and E. Stenberg, Principles of Biosensing with an Extended Coupling Matrix and Surface-Plasmon Resonance. Sensors and Actuators B-Chemical, 1993. 11(1-3): p. 63-72. 31. Jorgenson, R.C. and S.S. Yee, A Fiberoptic Chemical Sensor-Based on Surface-Plasmon Resonance. Sensors and Actuators B-Chemical, 1993. 12(3): p. 213-220. 32. Ebbesen, T.W., et al., Extraordinary optical transmission through sub-wavelength hole arrays. Nature, 1998. 391(6668): p. 667-669. 33. Ritchie, R.H., et al., Surface-Plasmon Resonance Effect in Grating Diffraction. Phys Rev Lett, 1968. 21(22): p. 1530-&. 34. Barnes, W.L., et al., Surface plasmon polaritons and their role in the enhanced transmission of light through periodic arrays of subwavelength holes in a metal film. Phys Rev Lett, 2004. 92(10): p. 107401. 35. Mayer, K.M. and J.H. Hafner, Localized Surface Plasmon Resonance Sensors. Chem Rev, 2011. 111(6): p. 3828-3857. 36. Henzie, J., M.H. Lee, and T.W. Odom, Multiscale patterning of plasmonic metamaterials. Nat Nanotechnol, 2007. 2(9): p. 549-54. 37. Liu, N., et al., Planar Metamaterial Analogue of Electromagnetically Induced Transparency for Plasmonic Sensing. Nano Lett, 2010. 10(4): p. 1103-1107. 38. Yonzon, C.R., et al., A comparative analysis of localized and propagating surface plasmon resonance sensors: the binding of concanavalin a to a monosaccharide functionalized self-assembled monolayer. J Am Chem Soc, 2004. 126(39): p. 12669-76. 39. Haes, A.J. and R.P. Van Duyne, A nanoscale optical biosensor: sensitivity and selectivity of an approach based on the localized surface plasmon resonance spectroscopy of triangular silver nanoparticles. J Am Chem Soc, 2002. 124(35): p. 10596-604. 40. Dahlin, A.B., J.O. Tegenfeldt, and F. Hook, Improving the instrumental resolution of sensors based on localized surface plasmon resonance. Anal Chem, 2006. 78(13): p. 4416-4423. 41. Homola, J., I. Koudela, and S.S. Yee, Surface plasmon resonance sensors based on

diffraction gratings and prism couplers: sensitivity comparison. Sensors and Actuators B- Chemical, 1999. 54(1-2): p. 16-24. 42. Ghosh, S.K., et al., Solvent and ligand effects on the localized surface plasmon resonance (LSPR) of gold colloids. Journal of Physical Chemistry B, 2004. 108(37): p. 13963- 13971. 43. Stewart, M.E., et al., Quantitative multispectral biosensing and 1D imaging using quasi- 3D plasmonic crystals. Proceedings of the National Academy of Sciences of the United States of America, 2006. 103(46): p. 17143-17148. 44. Yao, J.M., et al., Seeing molecules by eye: Surface plasmon resonance imaging at visible wavelengths with high spatial resolution and submonolayer sensitivity. Angewandte Chemie-International Edition, 2008. 47(27): p. 5013-5017. 45. Stewart, M.E., et al., Nanostructured plasmonic sensors. Chemical Reviews, 2008. 108(2): p. 494-521. 46. Yao, J., et al., Seeing molecules by eye: surface plasmon resonance imaging at visible wavelengths with high spatial resolution and submonolayer sensitivity. Angew Chem Int Ed Engl, 2008. 47(27): p. 5013-7. 47. Stewart, M.E., et al., Multispectral Thin Film Biosensing and Quantitative Imaging Using 3D Plasmonic Crystals. Analytical Chemistry, 2009. 81(15): p. 5980-5989. 48. Yao, J.M., et al., Functional Nanostructured Plasmonic Materials. Advanced Materials, 2010. 22(10): p. 1102-1110. 49. Yao, J.M., et al., Soft Embossing of Nanoscale Optical and Plasmonic Structures in Glass. Acs Nano, 2011. 5(7): p. 5763-5774. 50. Kauranen, M. and A.V. Zayats, Nonlinear plasmonics. Nature Photonics, 2012. 6(11): p. 737-748. 51. Maier, S.A., Plasmonics: Metal nanostructures for subwavelength photonic devices. Ieee Journal of Selected Topics in Quantum Electronics, 2006. 12(6): p. 1214-1220. 52. Talapin, D.V., et al., Prospects of Colloidal Nanocrystals for Electronic and Optoelectronic Applications. Chem Rev, 2010. 110(1): p. 389-458. 53. Yan, R.X., D. Gargas, and P.D. Yang, Nanowire photonics. Nature Photonics, 2009. 3(10): p. 569-576. 54. Hu, L.B., H. Wu, and Y. Cui, Metal nanogrids, nanowires, and nanofibers for transparent electrodes. Mrs Bulletin, 2011. 36(10): p. 760-765. 55. Gittins, D.I., et al., A nanometre-scale electronic switch consisting of a metal cluster and redox-addressable groups. Nature, 2000. 408(6808): p. 67-69. 56. Didiot, C., et al., Nanopatterning the electronic properties of gold surfaces with self- organized superlattices of metallic nanostructures. Nat Nanotechnol, 2007. 2(10): p. 617- 621. 57. Lal, S., S. Link, and N.J. Halas, Nano-optics from sensing to waveguiding. Nature Photonics, 2007. 1(11): p. 641-648. 58. Camden, J.P., et al., Controlled Plasmonic Nanostructures for Surface-Enhanced Spectroscopy and Sensing. Accounts of Chemical Research, 2008. 41(12): p. 1653-1661. 59. Schwartzberg, A.M. and J.Z. Zhang, Novel optical properties and emerging applications of metal nanostructures. Journal of Physical Chemistry C, 2008. 112(28): p. 10323- 10337. 60. Jain, P.K. and M.A. Ei-Sayed, Surface plasmon resonance sensitivity of metal nanostructures: Physical basis and universal scaling in metal nanoshells. Journal of

Physical Chemistry C, 2007. 111(47): p. 17451-17454. 61. Hao, F., et al., Symmetry Breaking in Plasmonic Nanocavities: Subradiant LSPR Sensing and a Tunable Fano Resonance. Nano Lett, 2008. 8(11): p. 3983-3988. 62. Zhang, J.Z. and C. Noguez, Plasmonic Optical Properties and Applications of Metal Nanostructures. Plasmonics, 2008. 3(4): p. 127-150. 63. Yu, B. and M. Meyyappan, Nanotechnology: Role in emerging nanoelectronics. Solid- State Electronics, 2006. 50(4): p. 536-544. 64. Biswas, A., et al., Advances in top-down and bottom-up surface nanofabrication: Techniques, applications & future prospects. Advances in Colloid and Interface Science, 2012. 170(1-2): p. 2-27. 65. Ronse, K., Optical lithography - a historical perspective. Comptes Rendus Physique, 2006. 7(8): p. 844-857. 66. Stevenson, J.T.M. and A.M. Gundlach, The Application of Photolithography to the Fabrication of Microcircuits. Journal of Physics E-Scientific Instruments, 1986. 19(9): p. 654-667. 67. Hicks, E.M., et al., Controlling plasmon line shapes through diffractive coupling in linear arrays of cylindrical nanoparticles fabricated by electron beam lithography. Nano Lett, 2005. 5(6): p. 1065-1070. 68. Tseng, A.A., et al., Electron beam lithography in nanoscale fabrication: recent development. Ieee Transactions on Electronics Packaging Manufacturing, 2003. 26(2): p. 141-149. 69. Tandon, U.S., An Overview of Ion-Beam Lithography for Nanofabrication. Vacuum, 1992. 43(3): p. 241-251. 70. Das, N., et al., Analysis of nano-grating-assisted light absorption enhancement in metal- semiconductor-metal photodetectors patterned using focused ion-beam lithography. Optics Communications, 2011. 284(6): p. 1694-1700. 71. Park, M., et al., Block copolymer lithography: Periodic arrays of similar to 10(11) holes in 1 square centimeter. Science, 1997. 276(5317): p. 1401-1404. 72. Cheng, J.Y., et al., Fabrication of nanostructures with long-range order using block copolymer lithography. Applied Physics Letters, 2002. 81(19): p. 3657-3659. 73. Whitesides, G.M., et al., Soft lithography in biology and biochemistry. Annual Review of Biomedical Engineering, 2001. 3: p. 335-373. 74. Xia, Y.N. and G.M. Whitesides, Soft lithography. Annual Review of Materials Science, 1998. 28: p. 153-184. 75. Kramer, S., R.R. Fuierer, and C.B. Gorman, Scanning probe lithography using self- assembled monolayers. Chem Rev, 2003. 103(11): p. 4367-4418. 76. Lu, W. and C.M. Lieber, Nanoelectronics from the bottom up. Nature Materials, 2007. 6(11): p. 841-850. 77. Knez, M., K. Niesch, and L. Niinisto, Synthesis and surface engineering of complex nanostructures by atomic layer deposition. Advanced Materials, 2007. 19(21): p. 3425- 3438. 78. Song, F., et al., Controllable synthesis and gas response of biomorphic SnO2 with architecture hierarchy of butterfly wings. Sensors and Actuators B-Chemical, 2010. 145(1): p. 39-45. 79. Przybyla, D.E. and J. Chmielewski, Higher-Order Assembly of Collagen Peptides into Nano- and Microscale Materials. Biochemistry, 2010. 49(21): p. 4411-4419.

80. Liu, J.F., et al., Metallization of Branched DNA Origami for Nanoelectronic Circuit Fabrication. ACS Nano, 2011. 5(3): p. 2240-2247. 81. Whitesides, G.M. and J.C. Love, The art of building small - Researchers are discovering cheap, efficient ways to make structures only a few billionths of a meter across. Scientific American, 2001. 285(3): p. 38-47. 82. May, G.S. and S.M. Sze, Fundamentals of semiconductor fabrication. Wiley international ed2004, New York: Wiley. xiii, 305 p. 83. Xiao, H., Introduction to semiconductor manufacturing technology2001, Upper Saddle River, N.J.: Prentice Hall. viii, 647 p. 84. Harriott, L.R., Limits of lithography. Proceedings of the Ieee, 2001. 89(3): p. 366-374. 85. Gates, B.D., et al., New approaches to nanofabrication: Molding, printing, and other techniques. Chem Rev, 2005. 105(4): p. 1171-1196. 86. Haynes, C.L. and R.P. Van Duyne, Nanosphere lithography: A versatile nanofabrication tool for studies of size-dependent nanoparticle optics. Journal of Physical Chemistry B, 2001. 105(24): p. 5599-5611. 87. Malinsky, M.D., et al., Nanosphere lithography: Effect of substrate on the localized surface plasmon resonance spectrum of silver nanoparticles. Journal of Physical Chemistry B, 2001. 105(12): p. 2343-2350. 88. Kelf, T.A., et al., Localized and delocalized plasmons in metallic nanovoids. Physical Review B, 2006. 74(24). 89. Rindzevicius, T., et al., Plasmonic sensing characteristics of single nanometric holes. Nano Lett, 2005. 5(11): p. 2335-2339. 90. Hanarp, P., et al., Control of nanoparticle film structure for colloidal lithography. Colloids and Surfaces a-Physicochemical and Engineering Aspects, 2003. 214(1-3): p. 23-36. 91. Aizpurua, J., et al., Optical properties of gold nanorings. Phys Rev Lett, 2003. 90(5). 92. Henzie, J., et al., Large-area nanoscale patterning: Chemistry meets fabrication. Accounts of Chemical Research, 2006. 39(4): p. 249-257. 93. Tang, C.B., et al., Evolution of block copolymer lithography to highly ordered square arrays. Science, 2008. 322(5900): p. 429-432. 94. Ginger, D.S., H. Zhang, and C.A. Mirkin, The evolution of dip-pen nanolithography. Angewandte Chemie-International Edition, 2004. 43(1): p. 30-45. 95. Salaita, K., Y.H. Wang, and C.A. Mirkin, Applications of dip-pen nanolithography. Nat Nanotechnol, 2007. 2(3): p. 145-155. 96. Weibel, D.B., W.R. DiLuzio, and G.M. Whitesides, Microfabrication meets microbiology. Nature Reviews Microbiology, 2007. 5(3): p. 209-218. 97. Rogers, J.A. and R.G. Nuzzo, Recent progress in soft lithography. Materials Today, 2005. 8(2): p. 50-56. 98. Delamarche, E., et al., Stability of molded polydimethylsiloxane microstructures. Advanced Materials, 1997. 9(9): p. 741-746. 99. Odom, T.W., et al., Improved pattern transfer in soft lithography using composite stamps. Langmuir, 2002. 18(13): p. 5314-5320. 100. Henzie, J., et al., Nanofabrication of Plasmonic Structures. Annu Rev Phys Chem, 2009. 60: p. 147-165. 101. Malyarchuk, V., et al., Spatially resolved biosensing with a molded plasmonic crystal. Applied Physics Letters, 2007. 90(20).

102. Baca, A.J., et al., Molded plasmonic crystals for detecting and spatially imaging surface bound species by surface-enhanced Raman scattering. Applied Physics Letters, 2009. 94(24). 103. Chang, S.H., S.K. Gray, and G.C. Schatz, Surface plasmon generation and light transmission by isolated nanoholes and arrays of nanoholes in thin metal films. Optics Express, 2005. 13(8): p. 3150-3165. 104. Yee, K.S., Numerical Solution of Initial Boundary Value Problems Involving Maxwells Equations in Isotropic Media. Ieee Transactions on Antennas and Propagation, 1966. Ap14(3): p. 302-&. 105. Taflove, A. and S.C. Hagness, Computational electrodynamics : the finite-difference time- domain method. 3rd ed. Artech House antennas and propagation library2005, Boston: Artech House. xxii, 1006 p. 106. Park, H.G., et al., Electrically driven single-cell photonic crystal laser. Science, 2004. 305(5689): p. 1444-1447. 107. Yanik, M.F., et al., All-optical transistor action with bistable switching in a photonic crystal cross-waveguide geometry. Opt Lett, 2003. 28(24): p. 2506-2508. 108. Wahl, P., et al., B-CALM: An open-source GPU-based 3D-FDTD with multi-pole dispersion for plasmonics. Optical and Quantum Electronics, 2012. 44(3-5): p. 285-290. 109. Maria, J., et al., Optimization of 3D Plasmonic Crystal Structures for Refractive Index Sensing. Journal of Physical Chemistry C, 2009. 113(24): p. 10493-10499. 110. Lezec, H.J. and T. Thio, Diffracted evanescent wave model for enhanced and suppressed optical transmission through subwavelength hole arrays. Optics Express, 2004. 12(16): p. 3629-3651. 111. Reilly, T.H., et al., Quantitative evaluation of plasmon enhanced Raman scattering from nanoaperture arrays. Journal of Physical Chemistry C, 2007. 111(4): p. 1689-1694. 112. Gao, H., et al., Rayleigh anomaly-surface plasmon polariton resonances in palladium and gold subwavelength hole arrays. Optics Express, 2009. 17(4): p. 2334-2340. 113. Stewart, M.E., et al., Multispectral thin film biosensing and quantitative imaging using 3D plasmonic crystals. Anal Chem, 2009. 81(15): p. 5980-9. 114. Yao, J., et al., Functional nanostructured plasmonic materials. Adv Mater, 2010. 22(10): p. 1102-10. 115. Yao, J., et al., Soft embossing of nanoscale optical and plasmonic structures in glass. ACS Nano, 2011. 5(7): p. 5763-74. 116. Mack, N.H., et al., Optical transduction of chemical forces. Nano Lett, 2007. 7(3): p. 733-737. 117. Riboh, J.C., et al., A nanoscale optical biosensor: Real-time immunoassay in physiological buffer enabled by improved nanoparticle adhesion. Journal of Physical Chemistry B, 2003. 107(8): p. 1772-1780. 118. Yu, C. and J. Irudayaraj, Quantitative evaluation of sensitivity and selectivity of multiplex nanoSPR biosensor assays. Biophys J, 2007. 93(10): p. 3684-3692. 119. Lin, L.M., et al., Surface plasmon resonance-based sensors to identify cis-regulatory elements. Anal Chem, 2004. 76(22): p. 6555-6559. 120. Huang, T.T., et al., Composite surface for blocking bacterial adsorption on protein biochips. Biotechnology and Bioengineering, 2003. 81(5): p. 618-624. 121. De Leebeeck, A., et al., On-chip surface-based detection with nanohole arrays. Anal Chem, 2007. 79(11): p. 4094-4100.

122. Tetz, K.A., L. Pang, and Y. Fainman, High-resolution surface plasmon resonance sensor based on linewidth-optimized nanohole array transmittance. Opt Lett, 2006. 31(10): p. 1528-1530. 123. Truong, T.T., et al., Nanopost plasmonic crystals. Nanotechnology, 2009. 20(43): p. 434011. 124. Mack, N.H., R. Dong, and R.G. Nuzzo, Quantitative imaging of protein adsorption on patterned organic thin-film arrays using secondary electron emission. J Am Chem Soc, 2006. 128(24): p. 7871-7881. 125. Raman, C.V., A new radiation (Reprinted from Indian J Phys, vol 2, pg 387-398, 1928). Current Science, 1998. 74(4): p. 382-386. 126. Nie, S.M. and S.R. Emery, Probing single molecules and single nanoparticles by surface- enhanced Raman scattering. Science, 1997. 275(5303): p. 1102-1106. 127. Stiles, P.L., et al., Surface-Enhanced Raman Spectroscopy. Annual Review of Analytical Chemistry, 2008. 1: p. 601-626. 128. Chanda, D., et al., Coupling of plasmonic and optical cavity modes in quasi-three- dimensional plasmonic crystals. Nature Communications, 2011. 2. 129. Haynes, C.L. and R.P. Van Duyne, Plasmon-sampled surface-enhanced Raman excitation spectroscopy. Journal of Physical Chemistry B, 2003. 107(30): p. 7426-7433. 130. Almeida, V.R., et al., All-optical control of light on a silicon chip. Nature, 2004. 431(7012): p. 1081-1084. 131. Reed, G.T., Device physics - The optical age of silicon. Nature, 2004. 427(6975): p. 595- 596. 132. Goetzberger, A. and C. Hebling, Photovoltaic materials, past, present, future. Solar Energy Materials and Solar Cells, 2000. 62(1-2): p. 1-19. 133. Muller, J., et al., TCO and light trapping in silicon thin film solar cells. Solar Energy, 2004. 77(6): p. 917-930. 134. Zhou, D.Y. and R. Biswas, Photonic crystal enhanced light-trapping in thin film solar cells. Journal of Applied Physics, 2008. 103(9). 135. Spinelli, P., et al., Plasmonic light trapping in thin-film Si solar cells. Journal of Optics, 2012. 14(2). 136. Papet, P., et al., Pyramidal texturing of silicon solar cell with TMAH chemical anisotropic etching. Solar Energy Materials and Solar Cells, 2006. 90(15): p. 2319-2328. 137. Sopori, B.L. and R.A. Pryor, Design of Anti-Reflection Coatings for Textured Silicon Solar-Cells. Solar Cells, 1983. 8(3): p. 249-261. 138. Campbell, P. and M.A. Green, Light Trapping Properties of Pyramidally Textured Surfaces. Journal of Applied Physics, 1987. 62(1): p. 243-249. 139. Ferry, V.E., et al., Light trapping in ultrathin plasmonic solar cells. Opt Express, 2010. 18(13): p. A237-A245. 140. Ferry, V.E., et al., Plasmonic Nanostructure Design for Efficient Light Coupling into Solar Cells. Nano Lett, 2008. 8(12): p. 4391-4397. 141. Pala, R.A., et al., Design of Plasmonic Thin-Film Solar Cells with Broadband Absorption Enhancements. Advanced Materials, 2009. 21(34): p. 3504-+. 142. Atwater, H.A. and A. Polman, Plasmonics for improved photovoltaic devices. Nature Materials, 2010. 9(3): p. 205-213. 143. Roh, S., T. Chung, and B. Lee, Overview of the Characteristics of Micro- and Nano- Structured Surface Plasmon Resonance Sensors. Sensors, 2011. 11(2): p. 1565-1588.

1.8 Figures

Figure 1.1 Schematic diagrams illustrating (a) a surface plasmon polariton (SPP) and (b) a localized surface plasmon resonance (LSPR). (Figure was reprinted and adapted with permission from [15]; copyright ©Annual Reviews, 2007.)

Figure 1.2 The dispersion curves for a surface plasmon polariton (SPP) mode (solid line) and free space photon (dot line). A SPP mode always has greater momentum than a free space photon of the same frequency. Therefore, the momentum mismatch should be overcome for coupling light and SPP. (Figure was reprinted and adapted with permission from [3]; copyright © Macmillan Publishers Ltd: Nature, 2003.)

Figure 1.4 (a) Schematic illustration of soft nanoimprint lithography protocol. (b) Optical image of an embossed plasmonic crystal. Hole depth is ~400 nm, hole diameters vary from ~0.24-1.06 μm, and hole spacings vary from ~0.50-1.74 μm. (c) SEM image of cylindrical nanohole array patterned by soft nanoimprint lithography (inset) individual hole and cross section view of nanohole. (Figure was reprinted and adapted with permission from [48]; copyright © John Wiley & Sons, Inc., 2010.)

Figure 1.5 (a) Experimental transmission spectrum (blue) for quasi-3D plasmonic crystal (~420 nm hole diameter, ~350 nm hole depth, ~720 nm hole spacing) and electrodynamic modeling transmission spectra with idealized gold distribution (green) or with isolated gold grains near edge of bottom gold disk (red). (b) Experimental transmission spectrum (black) for full-3D plasmonic crystal (~456 nm hole diameter, ~350 nm hole depth, ~748 nm hole spacing) and electrodynamic modeling transmission spectrum (red). (c, d) The calculated electric field plots corresponding to wavelengths for peaks marked B and C in the transmission spectra for (c) quasi-3D plasmonic crystal and (d) full-3D plasmonic crystal, respectively. (Figure was reprinted and adapted with permission from [48]; copyright © John Wiley & Sons, Inc., 2010.)

Figure 1.6 (a) Plot of normalized difference between transmission spectra for different refractive index environments in 1.4 wt% PEG solution (red line), 2.8 wt% PEG solution (green line), and 4.2 wt% PEG solution (blue line) and reference transmission spectrum in water. (b) Calculated sensitivity or figure of merit (FOM) for varying relief depth (solid black squares connected by black lines). Available experimental results for the relief depth dependence of plasmonic crystals with comparable configurations are also shown (solid red circles connected by red lines). (c) Calculated FOM for the plasmonic crystal assuming a uniform thickness gold layer covering the nanohole array. (d) Experimental FOM for varying the top gold layer thickness with the bottom and sidewall thicknesses being in proportions relative to the top layer. (Figure was reprinted and adapted with permission from [109]; copyright © American Chemical Society, 2009.)

Figure 1.7 Optical response of a quasi-3D plasmonic crystal to sequential injections of increasing concentrations of aqueous PEG solutions. (a) Color contour plot of the change in transmission (T) as a function of wavelength and time (with the corresponding injection sequence overlaid on the plot). (b) Integrated multispectral plasmonic response as a function of time. (Inset) A linear correlation to the change in refractive index. (Figure was reprinted and adapted with permission from [43]; copyright © National Academy of Sciences, 2006.)

Figure 1.8 (a) Side and front views of the flow cell used for the sensitivity calibration of the plasmonic crystal. Normal incidence transmission spectra were collected as a function of time as solutions of PEG were injected through the flow cell. (b) Digital picture of a plasmonic crystal mounted inside a flow cell. (Figure was reprinted and adapted with permission from [44]; copyright © John Wiley & Sons, Inc., 2006.)

Figure 1.9 (a) The normalized integrated response of a plasmonic crystal with immobilized antigoat IgG to increasing concentrations of goat IgG (red data points) and nonlinear regression of the binding curve (black line). (Figure was reprinted and adapted with permission from [47]; copyright © American Chemical Society, 2009.) (b) (Top) Color contour plot of the change in transmission of plasmonic crystal as a function of wavelength and time. The overlaid injection sequence corresponds to PBS (1), bBSA (2), BSA (3), and avidin (4). (Bottom) Integrated multispectral plasmonic response and corresponding effective thickness of the biotin-avidin- biotin assay (schematically illustrated in the inset). (Figure was reprinted and adapted with permission from [43]; copyright © National Academy of Sciences, 2006.) (c) (Top) Spectral sensitivity map consisting of difference spectra from crystal referenced to t = 0 s, after which the analyte solution is cycled between pH 7.86 and 1.44. (Bottom) The integrated plasmonic response corresponding to reversible changes in analyte solution from pH 7.86 to 1.44 (blue), 6.42 to 5.13 (red), and 5.76 to 5.66 (black). The smallest pH change (0.10) is well differentiated from the stable background signal (inset). (Figure was reprinted and adapted with permission from [116]; copyright © American Chemical Society, 2007.)

Figure 1.10 (a) A schematic illustrating the use of a three channel PDMS microfluidic device to pattern the surface of a crystal. (b) Transmitted light image of three proteins; (from left to right) fibrinogen (MW = 340 kDa), γ-globulins (MW = 160 kDa), and myoglobin (MW = 14.4 kDa). (c) The average line profile through the image in panel b. (Figure was reprinted and adapted with permission from [47]; copyright © American Chemical Society, 2009.)

Figure 1.11 (a) Transmitted-light image of ODT patterns (red features, the red boxes at the ends of each feature have dimensions on the order of 50 microns) on a plasmonic crystal. (b) Transmitted-light image of the sample after a 300 s of backfilling with HDT; same scale as (a). (c) Plot of image contrast (ODT intensity minus HDT intensity) as a function of the HDT backfilling time. (Figure was reprinted and adapted with permission from [44]; copyright © Wiley-VCH Verlag GmbH & Co., 2008.)

Figure 1.12 (a) SERS spectra of benzenethiol adsorbed onto areas of a molded SERS substrate composed of square lattices of cylindrical holes with different diameters: D=0.514 μm (blue); 0.689 μm (red); 1.06 μm (red). (b) Transmission intensity at 821 nm and SERS response as a function of diameters. (c) (i) Optical micrograph and (ii) SERS mapping of benzenethiol monolayer on nanohole array imprinted in previously photodefined SU-8. Scale bar is 5 μm. (Figure was reprinted and adapted with permission from [102]; copyright © American Institute of Physics, 2009.)

Figure 1.13 (a) Light trapping by scattering from metal nanoparticles at the surface of the solar cell. (b) Light trapping by the excitation of localized surface plasmons in metal nanoparticles embedded in the semiconductor. (c) Light trapping by the excitation of surface plasmon polaritons at the metal/semiconductor interface. (Figure was reprinted and adapted with permission from [142]; copyright © Macmillan Publishers Ltd: Nature Materials, 2010.)

Figure 1.14 (a) Schematic cross section of the patterned solar cell. Incident blue and red arrows indicate that blue light is absorbed before reaching the back contact while red light interacts more with the back patterns. (b) Photograph of finished imprinted patterned solar cell substrate. Each colored square is a separate device, with different particle diameter and pitch. (c) SEM of Ag overcoated patterns showing 290 nm diameter particles with 500 nm pitch. (d) Electrical measurements of plasmonic solar cell with 160 nm of a-Si:H. Curves are shown for square grid patterns of 250 nm diameter plasmonic scatterers at pitches of 500 nm and 700 nm, the flat reference cell, and the randomly textured Asahi cell. (e, f) External quantum efficiency spectra of nanopatterned and randomly textured solar cells obtained from (e) measurement and (f) simulation. (Figure was reprinted and adapted with permission from [139]; copyright © The Optical Society, 2010.)

CHAPTER 2

REFRACTIVE INDEX SENSING AND SURFACE-ENHANCED RAMAN SPECTROSCOPY USING SILVER-GOLD BIMETALLIC PLASMONIC CRYSTALS

2.1 Abstract

Herein we describe the fabrication and characterization of Ag and Au bimetallic plasmonic crystals as a system that exhibits improved capabilities for quantitative bulk- refractive-index (Bulk RI) sensing and surface-enhanced Raman spectroscopy (SERS) as compared to mono-metallic plasmonic crystals of similar form. The sensing optics, bimetallic plasmonic crystals consisting of sequential nanoscale layers of Ag coated by Au, are chemically stable and useful for quantitative forms of multispectral refractive index and spectroscopic chemical sensing. The results illustrate improvements in performance compared to previously reported homometallic devices that stem from distinctive plasmonic features and strong localized electric fields produced by Ag and Au layers optimized in terms of metal thicknesses and geometric features of the nanostructure. Finite-Difference Time-Domain (FDTD) simulations were used to theoretically verify the nature of the multimode plasmonic resonances generated by the devices and better understand the enhancements in multispectral refractive index (RI) and

SERS-based sensing that they offer. Taken together, the results demonstrate a robust and potentially low-cost platform for chemical/spectroscopic sensing.

2.2 Introduction

Studies of surface plasmons have attracted significant attention due to the diverse range of applications and processes in which they can be exploited. These include: laser emission[1];

44 light trapping[2, 3]; optical modulation[4]; and label-free means of chemical or biological sensing[5, 6]. Surface plasmons are collective oscillations of conduction electrons near metal surfaces that are excited by electromagnetic radiation incident at a metal/dielectric interface. This results in an evanescent electric field that can either extend from the metal surface to ~100-200 nm in length (surface plasmon polaritons), or as a localized resonance at the surface of a metal nanostructure (localized surface plasmons). The attributes of these excitations are highly sensitive to local refractive index changes, which in turn provide a capability for their exploitation in chemical and biological sensing.[7-9] In this way, surface plasmonic resonance

(SPR) sensors are promising as an analytical technique for real-time, fully label-free detection of molecules both quantitatively and qualitatively, as well as for monitoring surface interactions.[5,

10, 11] Surface-enhanced Raman spectroscopy, better known as SERS, is another important analytical application which utilizes the enhanced electromagnetic fields produced by surface plasmons.[12, 13] Raman scattering signals can be dramatically amplified on a plasmonic substrate, reportedly by as much as 10 to 11 orders of magnitude, reaching levels of sensitivity suitable for single molecule detection.[14, 15]

Generally speaking, photons cannot directly elicit plasmonic excitations on metal films from the air due to conservation requirements.[16, 17] To compensate for the mismatch in momentum between an incident photon and a plasmonic resonance, most studies to date have focused on metallic nanostructures such as nanoparticles,[18, 19] line gratings,[20] and nanoscale holes[21-26] or voids[27] to effect couplings and further obtain stronger electromagnetic fields and higher spatial resolution from localized surface plasmon resonances (LSPR). Many methods of fabrication have been described that provide structures capable of generating these plasmonic features.[28-33] Our work in this area has exploited soft nanoimprint lithography, a technique

45 that permits reproducible replication of precisely defined nano-sized features over a large area

(greater than 1 x 1 cm2),[34, 35] as a way to fabricate quantitative imaging-mode and multispectral plasmonic optics.[23, 24, 36] This fabrication method yields highly uniform arrays of nanoholes in a dielectric substrate that on metallization provide a plasmonic platform for SPR sensing and SERS.[22-26, 37, 38]

Noble metals such as Au and Ag are the most commonly used plasmonic materials because they generate strong plasmonic resonances at visible and near-infrared frequencies.[39]

Metallic Ag generates a stronger evanescent field and narrower plasmon resonances than does

Au that, in principle, are an advantage for both optical sensing and SERS.[40-42] It is Au, however, that is more widely utilized for analytical applications due to its long-term environmental stability, ease of surface modification, and biocompatibility.[43] The present study explores means for realizing synergies in the complementary attributes of each material, examining a bilayer/multi-metallic Ag/Au plasmonic crystal (PC) motif that is characterized by higher analytical sensitivities and stronger enhanced electric fields than our previously reported mono-metallic PC systems. Manipulating the composition of the thin metal films, their spatial distribution, and the design rules of the PC optical elements (in terms of the lattice parameter and features sizes of the nanoholes) is found to be an effective approach to optimizing responses in multispectral and SERS-based sensing. To illustrate the properties of these devices, the optical responses from exemplary PCs were acquired by 0th-order transmission measurements. Finite- difference time-domain (FDTD) calculations were performed to assist in characterizing how each system behaved in order to understand and obtain an optimized device form factor. The data illustrate that optimized Ag/Au bimetallic PCs harbor a useful potential for applications in chemical sensing.

2.3 Experimental Materials and Methods

2.3.1 Materials

The reagents were used as received without further purification unless otherwise specified. Spin-on-glass 315F (SOG) was purchased from Filmtronics and was filtered twice sequentially using 0.22 μm (Millipore) and 0.02 μm (Whatman Anotop 10) syringe filters immediately before use in order to remove nanoparticles in the SOG sol formed by hydrolysis.[26] Polydimethylsiloxane (soft PDMS; Sylgard 184, Dow corning) was made in a

10:1 ratio of PDMS base with curing agent. Hard PDMS components: poly(25-30% methylhydrosiloxane)-(dimethylsiloxane) (HMS-301), poly(7-8% vinylmethylsiloxane)-

(dimethylsiloxane), (VDT-731), Platinum divinyltetramethyldisiloxane (SIP6831.1) and (1,3,5,7- tetravinyl-1,3,5,7-tetramethylcyclotetrasiloxane) (7900) were purchased from Gelest.

Polyethylene glycol (PEG, MW = 10,000 g/mol) and benzenethiol (BT) (99.99 %) were purchased from Aldrich. 100% ethyl alcohol (Decon lab) was used to solvate benzenthiol.

Ultrapure water (18.2 MΩcm) was generated using a Millipore Milli-Q Academic A-10 system and used to prepare the PEG buffer (0-5.6 w%) solutions.

2.3.2 Plasmonic Crystal Fabrication via Soft Nanoimprint Lithography

The PCs were fabricated using soft nanoimprint lithography as previously reported.[22-

26, 37, 38] In brief, a composite hard-PDMS/soft-PDMS was cast on a patterned photoresist master with arrays of nanohole relief structures and used to fabricate the PCs. A glass slide was fully covered with liquid SOG, and spin cast (~950 rmp for 6 sec) to produce a thin and uniform layer of liquid SOG on the glass slide surface. The PDMS stamp was then pressed into the SOG coated glass slide and fastened in place to achieve conformal contact between the PDMS stamp

47 and SOG film. Before baking, the sample was left at room temperature for 7 min to facilitate evaporation of volatile organic components of the SOG material. The sample was soft baked at

110 °C for 5 min, followed by the careful removal the PDMS stamp. The embossed SOG film was then further cured at 200 °C for 5 min followed by baking at 160 °C overnight. Finally, the fully-cured SOG film was annealed at 450 °C under nitrogen for 1 h. The replicated SOG nanostructures consist of well-defined square arrays of nanoholes, patterned using sixteen different design rules that offered broad variations of response to optical frequencies. A ~5 nm titanium dioxide adhesion layer was deposited using atomic layer deposition (Cambridge nanotech) on the embossed SOG nanostructure followed by deposition of ~ 50 nm metallic film

(Au, Ag or both) via one of different methods. Sputter deposition in 5 mTorr argon (AJA

International) was utilized for substrates prepared for bulk-refractive-index (bulk IR) sensing experiments and electron beam (E-beam) evaporation (Temescal) was used for samples made for

SERS measurement. The schematic illustration of both SOG PC structures is given in Figure 2.1 and the scanning electron microscopy (SEM) images of nanohole arrays of different metal distributions are shown in Figure 2.2a (sputter deposition) and Figure 2.2b (E-beam evaporation).

The island-like metallic film structures (shown in inset of Figure 2.2a) were found to be susceptible over time to the penetration of liquid solution into the interfaces formed between the metal films and the SOG substrate. To prevent the degradation in performances this engendered, a conformal ~6 nm thick Al2O3 passivation film was deposited on top of metal by atomic layer deposition (Cambridge nanotech).

2.3.3 Bulk Refractive Index Sensing via Transmission-mode Spectroscopy

Transmission spectra in air and bulk RI dependent PC data were measured using a Varian

5G UV-Vis-NIR spectrophotometer with normal incident light and no temperature control. The

48 bulk RI measurements were carried out using methods that have previously reported.[22, 24, 26,

37, 44, 45] In this protocol, a PDMS flow cell was mounted on top of the PC and PEG buffer solutions of increasing concentration (pure water, 1.4 w%, 2.8 w%, 4.8 w%, and 5.6 w%) were injected into the PDMS flow cell with a syringe pump (Harvard Apparatus) at a flow rate of 0.1 mL/min. To change the solution, a new PEG solution was injected at a flow rate of 1.0 mL/min at the beginning stage for 2 min to completely flush the previous solution from the PDMS cell.

This was followed by injecting with a normal flow rate of 0.1 mL/min. Transmission spectra over a wavelength range of 355-1000 nm were collected throughout the process in order to monitor changes in multiple plasmonic responses to changes in surrounding dielectric environment.

2.3.4 Integrated Multispectral Response Calculation

In quantitative work, we consider both the position and intensity changes that occur in the plasmonic features present in the transmission spectrum over the entire range of collected wavelengths. The absolute difference values were computed using the differences between transmission spectra collected in each PEG solution and the spectrum recorded in pure water.

These values are subsequently integrated to account for wavelength-dependence of both negative and positive changes in spectroscopic intensity using Equation 2.1, a value having units of

%T nm .

Integrated response(% T ( ) d (2.1) wavelength

2.3.5 Surface-Enhanced Raman Spectroscopy (SERS) Measurements

A self-assembled monolayer of benzenethiol was deposited on top of the quasi-3D PCs by immersing them in a 15 mM benzenethiol ethanol solution for 12 hr, rinsing thoroughly with ethanol, and then drying with a stream of nitrogen gas. SERS measurements were made using a

SENTERRA dispersive Raman microscope (Bruker Optics) with an excitation laser wavelength of 785 nm, an excitation power of ~5 mW, focal length of 45 mm and acquisition time of 30 s.

The Raman spectra were collected over a Raman shift range of 500-1800 cm-1.

2.3.6 Finite-Difference Time-Domain (FDTD) Simulation of Plasmonic Nanostructures

A set of 3D FDTD simulations were used to model the normal incidence transmission spectra in air and water and the electromagnetic field distributions for full-3D PCs. The unit cell geometry was defined as an infinite square array of nanostructured holes on a metal film that are parallel to the x-y plane with a semi-infinite SOG material under the nanoholes and air above them. The unit cell spacing was 2 nm in all three dimensions with a total unit cell size of

NNNXXX  292  292  1200 grid points. The total simulation time for each unit cell was

100 fs. Perfectly matching uniaxial layers were applied on both sides of the z grid to avoid artificial reflection errors from the domain boundaries. Appropriate periodic boundary conditions were used to define the square array. The frequency-dependent Au[22] and Ag[46] permittivities are described by the Drude-Lorentzian model over a wavelength of 355-1500 nm. The dielectric constants for SOG and air were taken to be 1.43 and 1.00, respectively.

2.4 Results and Discussion

2.4.1 Structure and Properties of Bimetallic Plasmonic Crystals

Periodic nanostructures in metal films have been widely studied as a system to better understand the underlying physics of plasmon resonances as well as to develop high- performance platforms for a variety of SPR applications.[21, 47, 48] Efforts in this field have been directed particularly toward the unique optical properties of periodic nanostructures because of their capacities for extraordinary optical transmission, visual wavelength responsiveness, and high optical sensitivity to changes in the local environment.[49-51] In this study, we used square arrays of nanoholes molded into the surface of an inorganic SOG film using soft nanoimprint lithography. The SOG supported PC structure affords a chemically and thermally stable sub-wavelength optic for SPR sensing and for SERS measurements, as discussed in a previous publication.[26]

The SPRs produced using nanostructured metal films can be easily tuned by adjusting the geometric shape, thickness, and composition of the metal film, as well as the surrounding dielectric environment. Sputter coating and E-beam evaporation are both widely used tools for fabricating metal thin films, however the distribution of the deposited metal film resulting from each of the two methods are quite different due to the distinctive deposition characteristics. The step coverage of E-beam evaporation in particular is poor since it only offers an essentially unidirectional collision of source material atoms with the substrate, which has most of metal being deposited on the surface and bottom of the nanoholes. Sputtering, in contrast, is a less- directional deposition process that conformally coats the entire substrate. We have termed the metal nanostructures fabricated by E-beam deposition as quasi-3D structures because the metal

51 disk at the bottom of the nanohole is physically separated from the metal film on the top of the

SOG substrate, while the PCs with more continuous/conformal metal layers formed by sputter deposition are termed full-3D PCs.[37] Figure 2.1 schematically illustrates the metal film distribution differences resulting from fabrication using each of these two methods. The geometric differences of the metal films on the quasi-3D and the full-3D PCs were characterized using cross-section SEM imaging (inset of Figure 2.2). The top view SEM images of PCs in

Figure 2.2 demonstrate that PCs consist of uniform square arrays of nanoholes with periodicity of ~580 nm. Despite the fact that the diameters of embossed SOG nanostructures of both PCs are identical at ~380nm, the final diameter of nanoholes in the full-3D PC (Figure 2.2a) appears slightly smaller than that of the nanoholes present in the quasi-3D PC (Figure 2.2b) an effect that was attributed to differences in sidewall metal deposition resulting from differences in the deposition technique used.

In addition to changing the geometry of nanohole arrays and metal by altering the method of depositing the metal film, varying the metal film composition provides an additional means through which the optical responses of the PC can be modified. In this study, we used bimetallic layers to engineer the plasmonic responses of the device over a wider wavelength range, here exploiting the unique plasmonic modes of Ag in the visible range to supplement and enhance those provided by Au. This distinctive PC design further serves to compensate for the chemical instability of Ag by coating it with thin films of Au. To explore the variation in function resulting from the double layer film structure, we used five different ratios of Ag and Au while maintaining a constant overall metal film thickness on the SOG nanohole arrays. The ratios used were: 50 nm of Au (Au50), 10 nm of Ag and 40 nm of Au (Ag10Au40), 20 nm of Ag and 30 nm of Au (Ag20Au30), 30 nm of Ag and 20 nm of Au (Ag30Au20), and 40 nm of Ag and 10 nm of

Au (Ag40Au10). Figure 2.3 presents optical photographs of an entire series of PCs with varying compositions of Ag and Au. The sixteen squares (4×4 mm2) given on each plasmonic substrate exhibit distinctive colors due to light scattering from the nanohole arrays according to different design rules, with periodicities ranging from 0.49 to 1.75 μm and corresponding hole diameters ranging from 0.17 to 1.12 μm. The color changes seen from the typical color of Au to that of Ag follow an intuitive trend as the ratio of Ag in the metal film increases. Metrodological data developed using SEM demonstrated a high fidelity of the replicated structure to its imprint master, as illustrated in the nearly identical geometry (~340 nm of hole depth, ~380 nm of hole diameter, and ~580 nm of hole spacing) of an exemplary pair of PCs shown in Figure 2.2a (full-

3D) and 2.2b (quasi-3D).

The former structures are ones with particularly useful responses in the visible wavelength region of the optical spectrum. This is illustrated by the data present in Figure 2.4, which shows the 0th-order transmission spectra for six full-3D PCs recorded in air for varying ratios of Au and Ag. The variations in the transmission spectra can be attributed to the changes in the film compositions (Ag and Au thicknesses). The small discontinuity seen near 800 nm is an instrumental artifact, which results from a change of detectors during the scan. In Figure 2.4, the largest transmission peak, one spanning the near-infrared (NIR) range, is blue-shifted and its intensity decreases as the proportion of Ag in the metal film increases. Across the near UV- visible regions, enhanced transmission magnitudes and more spectroscopic features are observed as the mass-coverage of the Ag increases. The intensity of the characteristic peak for the Au surface plasmon resonance (at ~500 nm) is reduced as the proportion of Ag increases, while the peak intensity (at ~350 nm) corresponding to the bulk plasmon mode of Ag is enhanced. These qualitative trends are more quantitatively described by the results of theoretical modeling.

2.4.2 FDTD Modeling

Finite-difference time-domain (FDTD) calculations provide theoretical understandings of how the electromagnetic field interacts with the PC.[52-54] The transmission spectra of the periodic nanohole arrays in each of the PCs have multiple optical responses such as localized surface plasmon resonances (LSPRs)[28, 55, 56], Bloch wave surface plasmon polaritons (BW-

SPPs)[57, 58], Wood’s anomalies (WAs)[49, 59, 60], or a combination of these features[61].

Transmission spectra contributions also originate from the coupling of light directly transmitted through the metal film, as well as from the background metal absorption (at ~500 nm for Au and

~350 nm for Ag). The peaks in the transmission spectra, which correspond to BW-SPPs and

WAs, are correlated with the periodic structure of the nanoholes arrays, whereas LSPRs can be created by specific sub-wavelength-sized features of the metallic structures.

The appropriate optical constants for each material and the geometrical model of the periodic nanohole array are crucial for FDTD modeling because all of the plasmonic modes are very sensitive to the structural details and dielectric properties of the environment. The design rules for an optimized PC used in our calculations (see below), ones based on experimental metrological data were defined as a square array of nanoholes in a layer of SOG with diameters of 380 nm, depths of 340 nm, and center-to-center spacing of 580 nm, respectively. The dimensions of the metal film above the SOG nanohole arrays were as follows: a top metal layer of ~50 nm, bottom metal layer of ~20 nm, and sidewall metal layer of ~15 nm in thickness

(details are in Figure 2.5). The varying thickness values account for the shadowing effects that limit the degree of conformal coverage realized in the sputtering coating metallization step. In the calculation, we used a Drude plus two-pole Lorentzian model to obtain the dielectric constant of the metal as a function of wavelength.[46, 49]

Figure 2.6 presents the experimentally measured and calculated normal incidence transmission spectra and electric field distributions around the nanoholes for full-3D PCs with

Au50 (Figure 2.6a), Ag50 (Figure 2.6b), and Ag30Au20 (Figure 2.6c) mass-coverage metal films. The calculations and experimental data shown are for PCs resident in air, and assignments for specific features seen in the data are revealed by the calculations.

The wavelengths of the expected BW-SPPs and WAs can be predicted. BW-SPPs are standing waves corresponding to the coherent superposition of propagating SPPs.[57, 58] An approximate relation for the allowed wavelength of BW-SPPs on a nanohole array is

P ()     metal d (2.2) 22()   nnxy metal d

where P is the nanohole lattice spacing, metal ()is the wavelength dependent relative dielectric

constant for metal, and  d is the relative dielectric constant of the dielectric material that interfaces with the metal film. BW-SPPs of bimetallic plasmonic crystal are generated at the

air/gold interfaces ( d  1 for air) and the silver/SOG interfaces ( d 1.43 for SOG). The

solutions of Equation 2.2 excluding nnxy0are shown in Table 2.1, which yield the discrete

0th-order wavelength of the BW-SPPs. Table 2.1 also includes solutions for BW-SPPs wavelength for gold/SOG interfaces and air/silver interfaces for mono-layered plasmonic crystals

(Au50 and Ag50).

Table 2.1 Zero-order wavelength of the BW-SPPs (unit of wavelength is nm)

nx ny air/Au Au/SOG air/Ag Ag/SOG 0 1 611 860 602 857 308 303 346 0 2 331 353 415 370 484 1 1 415 642 449 629 305 356 1 2 316 - 384 453 310 361 2 2 - - 346 410

The Wood’s anomalies (WAs) are light waves diffracted to propagate in the plane of surface.[49, 59, 60] WAs act in the same fashion as BW-SPPs, but have a very narrow spatially extended nature without direct involvement of plasmonic responses. The condition for WAs is

   P d (2.3) 22 nnxy

Table 2.2 shows the wavelength of WAs at air/metal and metal/SOG interfaces.

Table 2.2 Wavelength of the BW-SPPs (unit of wavelength is nm)

air/metal metal/SOG 0 1 580.0000 693.5791 0 2 290.0000 346.7896 1 1 410.1219 490.4345 1 2 259.3839 310.1780 2 2 205.0609 245.2173

The largest peak (labeled C in Figure 2.6) is the result of strong LSPR excitations that are confined in the nanohole. Fano-like resonances, appearing as peak minima or maxima, are correlated with either BW-SPPs or WAs.[22, 61] From the theoretical analysis of BW-SPPs, we can verify that the peak minima labeled A in the transmission spectra originate from BW-SPPs.

At these wavelengths, the field intensity distributions seen near the interface of the top metal layer and air stand in good agreement with the characteristic field distribution expected for BW-

SPPs. Even so, many of the optical features observed in the transmission spectra are likely the result of complex interactions of light diffraction and concurrent plasmonic modes. The electric field distributions calculated for the wavelength at position B in each of the spectra are examples of features involving such couplings of multiple plasmonic modes. The intensity concentrated in the nanohole and near the sidewalls, for example, likely corresponds to LSPRs, and electric fields located at the metal-air interface are largely due to BW-SPPs and/or WAs.

The FDTD results (in Figure 2.6) are consistent with the experimental results in terms of the dependence of the 0th-order transmission spectra as a function of changes in the metal overlayer composition (presented in Figure 2.4). The data show, for example, that the peak intensities are reduced and the intensity of the transmission maxima (position C) are blue-shifted with increasing Ag content. The calculations further demonstrate several BW-SPPs features that characterize transmission for the Ag PCs at wavelength around 400 nm, a complexity in the blue region expected for Ag as compared with Au. The optical features appearing between 400-600 nm for the Ag PCs suggest a possibility for a highly sensitive response for such PCs at near-UV and visible wavelengths.[62, 63]

2.4.3 Bulk Refractive Sensitivity of Bimetallic Plasmonic Crystals

A single resonance peak analysis methodology does not provide a suitable means to determine the RI sensitivity of an imaging sensor. For this reason, we employed a multispectral protocol to quantify figures of merit (FOM) for sensitivity, as described in earlier reports on chemical sensing using PCs as an optic for imaging and spectroscopic detection.[22, 24, 26, 37]

Here we used a flow cell design the same as that previously reported to expose the PC to PEG solutions of varying concentrations to determine a FOM for its multispectral RI sensitivity.

Because sensitivities to bulk RI changes have been shown previously to be better for the full-3D

PC structure compared with quasi-3D counterpart,[37] the transmittance spectra for the former over a range of 355-1000 nm were collected as a function of time while PEG solutions of different concentration (from 1.4 w% to 5.6 w%) were passed through the flow cell. Integrated multispectral RI responses of five full-3D PCs with different mass-coverage metal films were measured to investigate and validate the best design rule for the PC for bulk RI sensing.

To determine an optimal nanohole array for RI sensing in the visible wavelength range before considering the effects of mass-fraction in metal thin film on plasmon resonances, we performed bulk RI sensing measurements (in Figure 2.7a) and FDTD calculations (in Figure 2.7b) using four design rules of nanoholes chosen from sixteen squares on a PC: 580 nm (p580), 780 nm (p780), 1100 nm (p1100), and 1600 nm (p1600) of periodicity. From the results of bulk solution-phase RI sensing (shown in Figure 2.7a), the most sensitive geometry for RI changes measured over a wavelength range of 300-800 nm was found to be a nanohole array with ~580 nm of periodicity and ~380 nm of hole diameter. The FDTD calculation for this structure

(presented in Figure 2.7b) confirms the physical origin of the bulk RI sensing results, in which the p580 Au50 full-3D plasmonic substrates generated the strongest plasmonic features at frequencies in the visible range (wavelengths spanning 600-800 nm). From the transmittance measurements made in air (shown in Figure 2.4), we can further infer that the shifts of peak maxima that occur as a result of changes made in the metal overlayer composition are small if nanohole geometries are identical. We therefore selected the p580 nanohole array as a suitable exemplar for the bulk RI sensitivity evaluation of bimetallic PCs carried out in this study.

Figure 2.8a shows the integrated responses of five full-3D PCs selected for a design rule giving optimal RI responses with different metal compositions, here plotted as a function of time.

Figure 2.8b demonstrates the linear integrated response changes for the Au50 and Ag10Au40 full-3D PC as a consequence of changes made in the bulk refractive index (RIU) of the contacting PEG solutions. The Ag10Au40 full-3D PC exhibits the largest RI sensitivity among the five PCs studied (the FOM values measured for these PCs are listed in Table 2.3). From 355-

1000nm, the RI sensitivity for the Ag10Au40 bimetallic PC (45,000 %/T nm RIU ) is nearly

~1.41 times that of the Au thin film PC (32,000 ). Higher mass-fractions of Ag in the bilayer stack progressively weaken the plasmonic response, as seen in the lower values of the

FOM.

Table 2.3 Figure of merit (FOM, %/T nm RIU ) for PCs with different ratio of Ag and Au layer

Wavelength (nm) Au50 Ag10Au40 Ag20Au30 Ag30Au20 Ag40Au10 355-1000 32,000 45,000 36,000 29,000 21,000

The transmission spectra measured in water provide an important qualitative insight into the varying of the bulk RI sensitivity of the PCs. We note that narrow and high intensity optical features are weighted significantly in the multispectral RI responses calculation Equation 2.1 as compared to broad and weak transmission features. Figure 2.9a shows transmission spectra of

PCs measured in water that illustrates these effects. Of particular note are the transmission intensity minima (appearing as a sharp feature near of 600 nm and 800 nm) and peak transmission maxima (a generally more complex feature appearing above 800 nm). One sees here that the full-3D Ag10Au40 PCs, the best system for bulk RI sensing, have the narrowest peak minima and the highest transmission feature (seen here at ~900 nm) as the qualitative

59 associations described above prediction. The FDTD results calculated for PCs immersed in water do not agree perfectly with the transmission spectra acquired experimentally because the mounted PDMS flow cell is not considered in the calculation. They do, however, still affirm the intuitive correlations noted above. Considering in this regard the calculated transmission spectra in water for the five PCs (shown in Figure 2.9b), one notes that, as predicted, the Ag10Au40 full-

3D PC presents both the narrowest peak minima and highest intensity in the calculated transmission.

2.4.4 SERS Enhancement of Bimetallic Plasmonic Crystals

Nanohole arrayed PCs are also promising as an easily replicable substrate for SERS. We have shown in an earlier report that plasmonic crystals formed on a molded SOG substrate provided excellent performance for Raman spectroscopic measurements.[26] In this study, we demonstrate that even greater Raman signal enhancements are afforded by using a bimetallic PC compared to those offered by a more conventional mono-metallic system. The total double metal layer thickness (50 nm) and the SOG nanohole array design rules (580 nm of periodicity and 380 nm of diameter) used for the SERS measurement are the same as those adopted in the RI measurements described above. Optimized thicknesses of Ag and Au metal films on the embossed SOG PC were tested using the quasi-3D PC device motif found to be a more suitable design for SERS measurement in earlier work.[25] The data present in Figure 2.10 experimentally confirmed that larger SERS enhancements do in fact result for the quasi-3D PC compared to the full-3D design for the bimetallic case as well. Table 2.4 shows the Raman intensities measured for each quasi-3D PC as a function of the different Au and Ag thickness ratios. As the data reveal, the best SERS enhancement are provided by the Ag40Au10 quasi-3D

PC, with a Raman intensity that was ~2.3 times higher than that of a comparable Au coated PC

(shown in Figure 2.11a).

Table 2.4 Raman intensity at 1073 cm-1 and 1573 cm-1 for five different full-3D PCs with 580 nm of periodicity.

Au50 Ag10Au40 Ag20Au30 Ag30Au20 Ag40Au10 1073 cm-1 3133 3395 4461 4845 7095 1573 cm-1 5499 5048 6437 7359 12830

The major Raman active vibrational modes of benzenethiol are located at 1073 cm-1 and

1574 cm-1 (which correspond to Raman shift peaks, relative to the 785 nm laser excitation, at 857 nm and 896 nm, respectively).[64] Past work has noted an approximate requirement for SERS enhancement by a PC is an enhanced electric field following at an optical frequency halfway

between the laser excitation wavelength ( ex , 785 nm) and the scattered wavelength of the

Raman active mode of interest.[25, 64, 65] This, then, corresponds to a case for the noted modes of benzenethiol where the PC provides maximal transmittance at wavelengths of 821 nm and 840

nm (the halfway point between and Raman ). The experimental transmission spectra of Au50 and Ag40Au10 quasi-3D PCs measured in air shown in Figure 2.11b qualitatively affirm this correlation between the magnitude of the SERS enhancement and the underlying optical properties of the plasmonic substrate.

These correlations are not fully predictive of optimal performances in SERS for a broader range of PC design rules. The larger data, for example, specifically show that the p580 PC is not the best design rule for SERS measurements and that the optical properties needed to obtain optimized performances are more complex and likely application-specific. We have found that measurements made using benzenethiol as an exemplary reporter molecule, for instance, signals

61 of a p780 quasi-3D Ag40Au10 PC are higher than those of the p580 quasi-3D Ag40Au10 PC, demonstrating an approximately ~1.3-1.8-fold larger enhancement (Figure 2.12a). The general correlation of the SERS intensity enhancement with specific features of the transmission intensity apparently still retain a useful qualitative character when one notes that the experimental results show higher transmission values at 820 nm and 841 nm for the p780 PC as compared with the p580 PC case (demonstrated in Figure 2.12b) We believe further work, and theoretical guidance, will be required to better understand the associations between optical properties and performance noted in these data.

2.5 Conclusions

The influences of the mass coverage of Ag and Au thin films as well as the geometrical parameters of the nanohole array design rules of bimetallic (bilayer) plasmonic crystals were experimentally studied through multispectral bulk RI sensing and SERS measurements conjoined with insights from theoretical modeling. This work demonstrates the feasibility of Ag/Au bimetallic PCs for analytical applications, further establishing that the bimetallic systems markedly surpass the quantitative performance of mono-metallic systems previously reported.

Through the use of FDTD calculations, we verify the nature of the different optical and plasmonic features that provide these attributes of performance. An optimized system was obtained through careful engineering of the system and analyzing its transmission spectra both experimentally and computationally. The methods described here hold additional value in that they provide guidance for further optimization of design rules of plasmonic devices to suit specific features of analytical applications. More directly, though, the multi-layered bimetallic

PC is found to offer great flexibility in terms of case of fabrication and features of performance, thereby establishing it as an interesting platform for highly sensitive forms of plasmon-based chemical sensing.

2.6 Acknowledgements

This work was supported by the U.S. Department of Energy under prime Award No. DE-

SC0001293 via subcontract 67N-1087758 from California Institute of Technology as part of the

DOE Energy Frontier Research Center on Light-Material Interactions in Energy Conversion and was performed using resources at the Frederick Seitz Materials Research Laboratory Central

Facilities at the University of Illinois, including the Center for Microanalysis of Materials; the

Micro/Nanofabrication Facility; and the Laser and Spectroscopy Facility. We also acknowledge the use of the Turing cluster, maintained and operated by the Computational Science and

Engineering Program at the University of Illinois.

2.7 References

1. Oulton, R.F., et al., Plasmon lasers at deep subwavelength scale. Nature, 2009. 461(7264): p. 629-632. 2. Pillai, S., et al., Surface plasmon enhanced silicon solar cells. Journal of Applied Physics, 2007. 101(9). 3. Corcoran, C.J., et al., Mechanisms of Enhanced Optical Absorption for Ultrathin Silicon Solar Microcells with an Integrated Nanostructured Backside Reflector. Acs Applied Materials & Interfaces, 2013. 5(10): p. 4239-4246. 4. Pacifici, D., H.J. Lezec, and H.A. Atwater, All-optical modulation by plasmonic excitation of CdSe quantum dots. Nature Photonics, 2007. 1(7): p. 402-406. 5. Homola, J., S.S. Yee, and G. Gauglitz, Surface plasmon resonance sensors: review. Sensors and Actuators B-Chemical, 1999. 54(1-2): p. 3-15. 6. Anker, J.N., et al., Biosensing with plasmonic nanosensors. Nature Materials, 2008. 7(6): p. 442-453. 7. Knoll, W., Interfaces and thin films as seen by bound electromagnetic waves. Annual Review of Physical Chemistry, 1998. 49: p. 569-638. 8. Barnes, W.L., A. Dereux, and T.W. Ebbesen, Surface plasmon subwavelength optics. Nature, 2003. 424(6950): p. 824-830. 9. Maier, S.A., Plasmonics : fundamentals and applications2007, New York: Springer. xxiv, 223 p. 10. Zhao, J., et al., Localized surface plasmon resonance biosensors. Nanomedicine, 2006. 1(2): p. 219-228. 11. Brockman, J.M., B.P. Nelson, and R.M. Corn, Surface plasmon resonance imaging measurements of ultrathin organic films. Annual Review of Physical Chemistry, 2000. 51: p. 41-63. 12. Campion, A. and P. Kambhampati, Surface-enhanced Raman scattering. Chemical Society Reviews, 1998. 27(4): p. 241-250. 13. Willets, K.A. and R.P. Van Duyne, Localized surface plasmon resonance spectroscopy and sensing. Annual Review of Physical Chemistry, 2007. 58: p. 267-297. 14. Camden, J.P., et al., Controlled Plasmonic Nanostructures for Surface-Enhanced Spectroscopy and Sensing. Accounts of Chemical Research, 2008. 41(12): p. 1653-1661. 15. Nie, S. and S.R. Emory, Probing Single Molecules and Single Nanoparticles by Surface- Enhanced Raman Scattering. Science, 1997. 275(5303): p. 1102-6. 16. Sambles, J.R., G.W. Bradbery, and F.Z. Yang, Optical-Excitation of Surface-Plasmons - an Introduction. Contemporary Physics, 1991. 32(3): p. 173-183. 17. Lezec, H.J., et al., Beaming light from a subwavelength aperture. Science, 2002. 297(5582): p. 820-822. 18. Murphy, C.J., et al., Anisotropic metal nanoparticles: Synthesis, assembly, and optical applications. Journal of Physical Chemistry B, 2005. 109(29): p. 13857-13870. 19. Kuwata, H., et al., Resonant light scattering from metal nanoparticles: Practical analysis beyond Rayleigh approximation. Applied Physics Letters, 2003. 83(22): p. 4625-4627. 20. Tan, W.C., et al., Flat surface-plasmon-polariton bands and resonant optical absorption on short-pitch metal gratings. Physical Review B, 1999. 59(19): p. 12661-12666. 21. Ebbesen, T.W., et al., Extraordinary optical transmission through sub-wavelength hole arrays. Nature, 1998. 391(6668): p. 667-669.

22. Stewart, M.E., et al., Quantitative multispectral biosensing and 1D imaging using quasi- 3D plasmonic crystals. Proceedings of the National Academy of Sciences of the United States of America, 2006. 103(46): p. 17143-17148. 23. Stewart, M.E., et al., Multispectral Thin Film Biosensing and Quantitative Imaging Using 3D Plasmonic Crystals. Analytical Chemistry, 2009. 81(15): p. 5980-5989. 24. Yao, J.M., et al., Seeing molecules by eye: Surface plasmon resonance imaging at visible wavelengths with high spatial resolution and submonolayer sensitivity. Angewandte Chemie-International Edition, 2008. 47(27): p. 5013-5017. 25. Baca, A.J., et al., Molded plasmonic crystals for detecting and spatially imaging surface bound species by surface-enhanced Raman scattering. Applied Physics Letters, 2009. 94(24). 26. Yao, J.M., et al., Soft Embossing of Nanoscale Optical and Plasmonic Structures in Glass. Acs Nano, 2011. 5(7): p. 5763-5774. 27. Cole, R.M., et al., Understanding plasmons in nanoscale voids. Nano Letters, 2007. 7(7): p. 2094-2100. 28. Prikulis, J., et al., Optical spectroscopy of nanometric holes in thin gold films. Nano Letters, 2004. 4(6): p. 1003-1007. 29. Haynes, C.L. and R.P. Van Duyne, Nanosphere lithography: A versatile nanofabrication tool for studies of size-dependent nanoparticle optics. Journal of Physical Chemistry B, 2001. 105(24): p. 5599-5611. 30. Malinsky, M.D., et al., Nanosphere lithography: Effect of substrate on the localized surface plasmon resonance spectrum of silver nanoparticles. Journal of Physical Chemistry B, 2001. 105(12): p. 2343-2350. 31. Aizpurua, J., et al., Optical properties of gold nanorings. Physical Review Letters, 2003. 90(5). 32. Rindzevicius, T., et al., Plasmonic sensing characteristics of single nanometric holes. Nano Letters, 2005. 5(11): p. 2335-2339. 33. Hanarp, P., M. Kall, and D.S. Sutherland, Optical properties of short range ordered arrays of nanometer gold disks prepared by colloidal lithography. Journal of Physical Chemistry B, 2003. 107(24): p. 5768-5772. 34. Gates, B.D., et al., New approaches to nanofabrication: Molding, printing, and other techniques. Chemical Reviews, 2005. 105(4): p. 1171-1196. 35. Ganesan, R., et al., Direct nanoimprint lithography of Al2O3 using a chelated monomer- based precursor. Nanotechnology, 2012. 23(31). 36. Le, A.P., et al., Quantitative Reflection Imaging of Fixed Aplysia californica Pedal Ganglion Neurons on Nanostructured Plasmonic Crystals. Journal of Physical Chemistry B, 2013. 117(42): p. 13069-13081. 37. Yao, J.M., et al., Functional Nanostructured Plasmonic Materials. Advanced Materials, 2010. 22(10): p. 1102-1110. 38. Truong, T.T., et al., Nanopost plasmonic crystals. Nanotechnology, 2009. 20(43). 39. Johnson, P.B. and R.W. Christy, Optical Constants of Noble Metals. Physical Review B, 1972. 6(12): p. 4370-4379. 40. Zynio, S.A., et al., Bimetallic layers increase sensitivity of affinity sensors based on surface plasmon resonance. Sensors, 2002. 2(2): p. 62-70.

41. Freeman, R.G., et al., Ag-clad Au nanoparticles: Novel aggregation, optical, and surface- enhanced Raman scattering properties. Journal of Physical Chemistry, 1996. 100(2): p. 718-724. 42. Liu, C.H., et al., Bimetallic structure fabricated by laser interference lithography for tuning surface plasmon resonance. Optics Express, 2008. 16(14): p. 10701-10709. 43. Wang, Y.L., et al., Surface-enhanced Raman scattering of silver-gold bimetallic nanostructures with hollow interiors. Journal of Chemical Physics, 2006. 125(4). 44. Maria, J., et al., Optimization of 3D Plasmonic Crystal Structures for Refractive Index Sensing. Journal of Physical Chemistry C, 2009. 113(24): p. 10493-10499. 45. Stewart, M.E., et al., Nanostructured plasmonic sensors. Chemical Reviews, 2008. 108(2): p. 494-521. 46. Lee, T.W. and S.K. Gray, Subwavelength light bending by metal slit structures. Optics Express, 2005. 13(24): p. 9652-9659. 47. Ghaemi, H.F., et al., Surface plasmons enhance optical transmission through subwavelength holes. Physical Review B, 1998. 58(11): p. 6779-6782. 48. Porto, J.A., F.J. Garcia-Vidal, and J.B. Pendry, Transmission resonances on metallic gratings with very narrow slits. Physical Review Letters, 1999. 83(14): p. 2845-2848. 49. Lezec, H.J. and T. Thio, Diffracted evanescent wave model for enhanced and suppressed optical transmission through subwavelength hole arrays. Optics Express, 2004. 12(16): p. 3629-3651. 50. Reilly, T.H., et al., Quantitative evaluation of plasmon enhanced Raman scattering from nanoaperture arrays. Journal of Physical Chemistry C, 2007. 111(4): p. 1689-1694. 51. Gao, H., et al., Rayleigh anomaly-surface plasmon polariton resonances in palladium and gold subwavelength hole arrays. Optics Express, 2009. 17(4): p. 2334-2340. 52. Taflove, A. and S.C. Hagness, Computational electrodynamics : the finite-difference time-domain method. 3rd ed. Artech House antennas and propagation library2005, Boston: Artech House. xxii, 1006 p. 53. Maier, S.A., et al., Experimental demonstration of fiber-accessible metal nanoparticle plasmon waveguides for planar energy guiding and sensing. Applied Physics Letters, 2005. 86(7). 54. Artar, A., A.A. Yanik, and H. Altug, Fabry-Peacuterot nanocavities in multilayered plasmonic crystals for enhanced biosensing. Applied Physics Letters, 2009. 95(5). 55. Wannemacher, R., Plasmon-supported transmission of light through nanometric holes in metallic thin films. Optics Communications, 2001. 195(1-4): p. 107-118. 56. Degiron, A., et al., Optical transmission properties of a single subwavelength aperture in a real metal. Optics Communications, 2004. 239(1-3): p. 61-66. 57. Salomon, L., et al., Near-field distribution of optical transmission of periodic subwavelength holes in a metal film. Physical Review Letters, 2001. 86(6): p. 1110-1113. 58. Darmanyan, S.A. and A.V. Zayats, Light tunneling via resonant surface plasmon polariton states and the enhanced transmission of periodically nanostructured metal films: An analytical study. Physical Review B, 2003. 67(3). 59. Treacy, M.M.J., Dynamical diffraction in metallic optical gratings. Applied Physics Letters, 1999. 75(5): p. 606-608. 60. Sarrazin, M., J.P. Vigneron, and J.M. Vigoureux, Role of Wood anomalies in optical properties of thin metallic films with a bidimensional array of subwavelength holes. Physical Review B, 2003. 67(8).

61. Chang, S.H., S.K. Gray, and G.C. Schatz, Surface plasmon generation and light transmission by isolated nanoholes and arrays of nanoholes in thin metal films. Optics Express, 2005. 13(8): p. 3150-3165. 62. Chen, H.J., et al., Shape- and size-dependent refractive index sensitivity of gold nanoparticles. Langmuir, 2008. 24(10): p. 5233-5237. 63. Sagle, L.B., et al., Advances in localized surface plasmon resonance spectroscopy biosensing. Nanomedicine, 2011. 6(8): p. 1447-1462. 64. Chanda, D., et al., Coupling of plasmonic and optical cavity modes in quasi-three- dimensional plasmonic crystals. Nature Communications, 2011. 2. 65. Haynes, C.L. and R.P. Van Duyne, Plasmon-sampled surface-enhanced Raman excitation spectroscopy. Journal of Physical Chemistry B, 2003. 107(30): p. 7426-7433.

2.8 Figures

Figure 2.1 3D and 2D schematic illustration of the full-3D and quasi-3D PCs. To fabricate metal films, sputtering was used for the full-3D PC and E-beam evaporating was used for quasi-3D PC.

Figure 2.2 SEM images of (a) full-3D and (b) quasi-3D PC with diameters and periodicities of ~380 and 580 nm. The scale bar corresponds to 5 μm. Inset: cross-sectional SEM image of individual nanohole. The scale bar corresponds to 100 nm.

Figure 2.3 Optical image of an embossed PCs with different compositions of metal layers: ~50 nm Au (Au50), ~10 nm Ag and ~40 nm Au (Ag10Au40), ~ 20 nm Ag and ~30 nm Au (Ag20Au30), ~30 nm Ag and ~20 nm Au (Ag30Au20), ~40 nm Ag and ~10 nm Au (Ag40Au10), ~50 nm Ag (Ag50). Hole depth is ~340 nm, and hole spacing varies from ~500 nm to ~1740 nm.

Figure 2.4 Normal incident transmission spectra of five full-3D PCs with different metal composition.

Figure 2.5 Schematic illustration of an individual nanohole on a full 3D plasmonic crystal used for FDTD modeling. For simplicity, we assumed that metal films are uniformly distributed on a cylindrical SOG nanohole with a right angle edge. The full 3D plasmonic crystal with the periodicity (P) of hole of 580 nm, hole depth (H) of 340 nm, and diameter of hole (D) of 380 nm are used for all FDTD calculation. The sidewall thickness (Tside) for Ag and Au mono-metallic plasmonic crystals used in these calculations were slightly different (16 nm for Ag sidewall and 12 nm for Au sidewall), while thicknesses of top and bottom layers (Ttop=50 nm and Tbottom= 20 nm) for both plasmonic metal layer were same. We only monitored top layer thickness of metal to control thickness of metallic resonant layer without consideration of sidewall thickness.

Figure 2.6 Experimental transmission spectra (black), electrodynamic modeling transmission spectra (red), and 2D calculated electric field plots corresponding to wavelengths for peaks labeled as A, B, and C in the transmission plot of (a) Au50, (b) Ag50 and (c) Ag30Au20 full-3D PCs (~380 nm hole diameter, ~ 340 nm relief depth, ~ 580 nm hole spacing, ~50 nm top metal layer, ~20 nm metal sidewall, and ~15 nm bottom metal layer)

Figure 2.7 (a) Integrated multispectral responses of full 3D plasmonic crystals with different periodicity of nanohole arrays (~ 580 nm, ~780 nm, ~1100 nm, and ~1600 nm periodicities) to injections of increasing concentration of aqueous PEG solutions (from 1.4 w% to 5.6 w%). Transmittance changes as a function of time were collected over a wavelength range of 300-800 nm. (b) Normal incident transmission spectra of full 3D plasmonic crystals with different periodicity of nanohole arrays used in bulk refractive index sensing.

Figure 2.8 (a) Integrated multispectral responses of five full-3D PCs with different metal composition to injections of increasing concentration of aqueous PEG solutions (from 1.4 w% to 5.6 w%). Transmittance changes as a function of time were collected over a wavelength range of 355-1000 nm. (b) Linear change in average integrated response with refractive index change on Au50 and Ag10Au40 full-3D PCs.

Figure 2.9 (a) Left: experimental transmission spectra in water of five full 3D plasmonic crystals with different metal compositions. Right: Magnified view of the transmission spectra of Au50 and Ag10Au40 full 3D plasmonic crystals over a wavelength range of 500-1000 nm. (b) FDTD calculated transmission spectra in water of the full 3D plasmonic crystals with five different metal compositions. Right: Magnified view of the transmission spectra of Au50 and Ag10Au40 full 3D plasmonic crystals over a wavelength range of 500-1000 nm.

Figure 2.10 (a) SERS spectra of benzenethiol adsorbed onto full 3D and quasi 3D Ag40Au10 plasmonic crystals measured with laser power of ~5 mW. (b) SERS spectra of benzenethiol adsorbed onto full 3D and quasi 3D Au50 plasmonic crystals measured with laser power of ~5 mW.

Figure 2.11 (a) SERS spectra of benzenethiol adsorbed onto Au50 and Ag40Au10 quasi-3D PCs measured with laser power of ~5 mW. (b) Experimental transmission spectra in air for Au50 and Ag40Au10 quasi-3D PCs used to collect the Raman spectra in (a). The locations of the laser excitation and Raman peaks are marked as ex , Raman1 , and Raman2 , respectively.

Figure 2.12 (a) SERS spectra of benzenethiol adsorbed onto Ag40Au10 quasi 3D plasmonic crystals with ~580 nm and ~780 nm of periodicity measured with laser power of ~5 mW. (b) Experimental transmission spectra in air for Ag40Au10 quasi 3D plasmonic crystals with ~580 nm and ~780 nm periodicity used to collect the Raman spectra in (a). The locations of the laser excitation and Raman peaks are marked as ex , Raman1 , and Raman2 , respectively.

CHAPTER 3

QUANTITATIVE REFLECTION IMAGING OF FIXED APLYSIA CALIFORNICA PEDAL GANGLION NEURONS ON NANOSTRUCTURED PLASMONIC CRYSTALS

The majority of the text and figures included here is reproduced with permission from the previously published paper: A-P. Le, S. Kang, L. B. Thompson, S. S. Rubakhin, J. V. Sweedler, J. A. Rogers, R. G. Nuzzo, “Quantitative Reflection Imaging of Fixed Aplysia californica Pedal Ganglion Neurons on Nanostructured Plasmonic Crystals”, J. Phys. Chem. B., 2013, 117 (42), 13069-13081. Copyright 2013, American Chemical Society.

3.1 Abstract

Studies of the interactions between cells and surrounding environment and their responses to distinct chemical and physical cues are essential to understanding the regulation of cell growth, migration, and differentiation. In this work, we demonstrate capability of label-free optical imaging technique - surface plasmon resonances (SPR) - to investigate quantitatively the relative thickness of complex biomolecular structures using a nanoimprinted plasmonic crystal and laboratory microscope. Polyelectrolyte films of different thicknesses deposited by layer-by- layer assembly served as the model system to calibrate the reflection contrast response originating from SPRs. The calibrated SPR system allowed quantitative analysis of the thickness of region of the interface between the cell culture substrate and cellular membrane regions of fixed Aplysia californica pedal ganglion neurons. Bandpass filters were used to isolate spectral regions of reflected light with distinctive image contrast changes. Combining of the data from images acquired using different bandpass filters leads to increase image contrast and sensitivity to topological differences in interface thicknesses. This SPR-based imaging technique is restricted in measurable thickness range (~100-200 nm) due to the limited plasmonic sensing volume, but we complement this technique with an interferometric analysis method. Described here simple reflection imaging techniques show promise as quantitative methods for analyzing

80 surface thicknesses at nanometer scale over large areas in real-time and in physicochemical diverse environments.

3.2 Introduction

Quantitative investigations of cell structure and cell-substrate interactions are important in understanding mechanisms and dynamics of cell growth.[1-7] Two-dimensional optical imaging is widely performed to study structural and biological processes of cells. Bright field microscopy is a standard and extensively used optical imaging technique. It has the limitation, however, of low imaging contrast and resolution.[8] Phase contrast microscopy (PCM)[9] and differential interference contrast microscopy[10, 11] have been developed to increase the contrast and resolution of images without cell staining with different dyes. Unfortunately, both methods create distortions around the edges of the sample due to halo effects[12] and require complicated adjustments to the microscope’s condenser and phase contrast. Fluorescence microscopy is a tremendously important technique — one well suited to studying specific structures or processes within cells by detecting naturally fluorescent compounds or tagging molecules of interest with fluorophores such as small organic dyes or quantum dots.[5, 13, 14] Through careful choice of fluorophores, multiple structures can be tagged and visualized simultaneously, but a primary difficulty of using fluorescent labels is their introduction and detection without altering the physiological and morphological parameters of live cells.[15-17]

While these two-dimensional imaging techniques provide useful information about cells, their data is ultimately a projection of a three-dimensional image with a deficiency of information along the z axis. Scanning confocal microscopy[5, 18] and spinning disk confocal microscopy[19, 20] lift this constraint and allow creation 3D images of cells by acquiring in-

81 focus 2D optical slices from several depths within a sample and then reconstructing a 3D composite image. These methods, while extremely powerful, have inherent limitations that stem from diffraction and fluorescence saturation, and protocols for fluorescently labeling cellular structures often require extensive optimization.[21] Digital holographic microscopy provides label-free and quantitative phase contrast images by digitally recording the light wavefront from the object under investigation as a hologram and calculating the object images using a numerical reconstruction algorithm.[22] Spatial light interference microscopy combines PCM and holography to measure structures and dynamics in cells via interferometry: PCM is able to measure the intrinsic contrast and holography renders quantitative phase maps.[23] While these techniques are capable of three-dimensional imaging, they require complex and highly demanding setups and data processing.

An alternative to optical imaging systems, atomic force microscopy (AFM) can be used to directly map the topography of a surface with feature heights ranging from angstroms to micrometers.[24-26] The relatively long acquisition time needed to collect AFM data is, however, a major drawback to monitoring dynamics in real time. Live cells, which are soft, can also be physically damaged during data acquisition due to physical contact between the moving

AFM tip and the sample. More significantly, though, the interfaces formed by cell-substrate contacts are generally inaccessible to study by AFM.

Surface plasmon resonance (SPR) imaging is a label-free and highly sensitive method for studying interactions occurring between a substrate and a biological sample. SPRs are coherent oscillations of electron density at the interface of a metal and a dielectric, which generate an evanescent electric field that decays exponentially within ~100-200 nm from the surface of the metal.[27-30] The optical sensitivity of SPR to changes in refractive index and layer thickness

82 within the sensing volume has been exploited in biological-sensing, providing both qualitative and quantitative data in real-time without requirements for external labeling.[31-35] It is particularly well suited for studying the effect of substrate interactions on cellular morphology and growth. SPR-imaging studies have demonstrated both refractive index mapping and thickness profiling of cell-surface contacts.[36-38] Many current SPR-imaging techniques, however, require relatively complex optical alignment schemes because they utilize a prism to couple polarized light into a flat gold film.[27, 31, 35, 39]

Nanostructured grating-based systems represent, in principle, a simple platform design for quantitative SPR-imaging, one capable of high spatial resolution arising from the excitation of localized surface plasmon resonances (LSPRs).[29, 40-45] We have previously demonstrated plasmonic imaging with molecular scale sensitivity in white light using nanohole plasmonic crystals fabricated by nanoimprint soft lithography, a common laboratory microscope, and a charge-coupled device (CCD) camera.[46] The quantitative optical responses of this plasmonic crystal system in air as a function of thickness (up to ~90 nm thick of material immediately adjacent to metal surface) were characterized in these earlier studies.[42]

We extend this method in the present work to the imaging of complex biological surface structures, explicitly treating within FDTD theoretical simulations the use of the wavelength dependence as a means to enhance image contrasts and film thickness sensitivities via the quantitative analysis of multispectral plasmonic reflection images collected using bandpass filters. We carried out companion experimental studies using a model full 3D plasmonic crystal with high analytical sensitivity for quantitative imaging in white light, enhancing its responses using multispectral measurements. The multispectral image contrasts were calibrated against polyelectrolyte layer-by-layer assemblies with well-defined thicknesses and applied to quantitate

83 the thicknesses of peripheral structures of Aplysia pedal neurons cultured and fixed on the surface of a nanostructured plasmonic crystal. We further describe how non-plasmonic optical effects revealed in these multispectral images can be used to extend the described quantitative optical measurements to structural features of a cell that are thicker than the plasmonic sensing volume.

3.3 Experimental Materials and Methods

3.3.1 Materials

Reagents were used as received without further purification unless otherwise specified.

Spin-On-Glass (Accuglass 314) was purchased from Honeywell and was filtered through 0.02

μm syringe filters (Whatman Anotop 10) immediately before use. Photocurable polyurethane

(NOA 73) was purchased from Norland Products. Polydimethylsiloxane (soft PDMS; Sylgard

184, Dow corning) was made in a 10:1 ratio of PDMS base with curing agent. Hard PDMS components: poly(25-30% methylhydrosiloxane)-(dimethylsiloxane) (HMS-301), poly(7-8% vinylmethylsiloxane)-(dimethylsiloxane), (VDT-731), Platinum divinyltetramethyldisiloxane

(SIP6831.1) and (1,3,5,7-tetravinyl-1,3,5,7-tetramethylcyclotetrasiloxane) (7900) were purchased from Gelest. Manganese chloride tetrahydrate, poly(sodium 4-styrenesulfonate ) (PSS,

MW = 70,000 g/mol), poly(allylamine hydrochloride) (PAH, MW = 70,000 g/mol), and 4,4’- dithiodibutyric acid (DTBA) were purchased from Sigma-Aldrich. Ultrapure water (18 MΩ) was generated using a Millipore Milli-Q Academic A-10 system and used to prepare the polyelectrolyte solutions.

3.3.2 Plasmonic Crystal Fabrication via Soft Nanoimprint Lithography

Full 3D plasmonic crystals were fabricated using soft nanoimprint lithography as

84 previously reported.[29, 41-45] Briefly, a composite hard-PDMS/soft-PDMS was cast from a patterned photoresist master with an array of nanohole relief structures and used to mold a replica master in spin-on-glass (SOG). An additional composite PDMS stamp was then cast from the SOG master and subsequently used to fabricate the plasmonic crystals used in this study; a schematic illustration of the nanoimprinting process is presented in Figure 3.1. A photocurable polyurethane polymer (NOA) was drop cast onto a glass side, and the second PDMS stamp was pressed into the liquid pre-polymer and then exposed to ultraviolet light (UVOCS UV-ozone cleaning chamber) for 5 min to cure the polyurethane. The stamp was peeled away from the patterned polymer structure, and the embossed polymer was cured at 65 °C overnight. The replicated nanostructures were a well-defined square array of nanowells with a hole spacing

(center to center) of ~740 nm, a hole diameter of ~460 nm, and a relief depth of ~300 nm (SEM image of nanoholes were shown in Figure 3.2a). An adhesion layer (5 nm TiO2) and a metal layer

(32 nm gold) were sputtered onto the relief structure (AJA international, 5 mTorr argon) to complete the plasmonic crystal fabrication.

3.3.3 Growth of Polyelectrolyte Layer-by-Layer Assemblies on Gold Films

Polyelectrolyte layer-by-layer assemblies were grown on the surface of the plasmonic crystals and on gold-coated silicon wafers following previously reported procedures.[42, 47] A small drop of NOA was applied to one corner of the plasmonic crystal and cured to block a region of the plasmonic crystal from adsorption of the polyelectrolyte film. Carboxyl-terminated self-assembled monolayers were formed on the gold film surface by immersing the substrate in ethanolic solutions of DTBA (33 mM) for 24 h, after which the gold films were rinsed thoroughly with ethanol and dried with nitrogen gas. The thiol end-group of DTBA helped to anchor itself on the gold surface and the negatively charged carboxylic acid end-group served as

85 the initial layer to electrostatically attract the positively charged PAH. Layers of PAH and PSS were alternately adsorbed onto the surface to build up the polyelectrolyte assembly. The substrates were immersed in a PAH solution (3 mg PAH/mL in water, pH = 8.0) for 5 min, rinsed thoroughly with water and dried with nitrogen gas. The substrate was then immersed in a PSS solution (3 mg PSS/mL in 1M MnCl2, pH = 2.0) for 90 s, rinsed with water and dried with nitrogen gas. Each round of polyelectrolyte deposition consisted of the adsorption of one PAH and one PSS layer, and reflection images and ellipsometry measurements were collected after each round of polyelectrolyte depositions.

3.3.4 Ellipsometry of Polyelectrolyte Layer-by -Layer (LBL) Assemblies

The thicknesses of polyelectrolyte LBL assemblies grown on gold-coated pieces of silicon were measured using a Woollam VASE spectroscopic ellipsometer using 50° and 70° incident angles; data were collected over a wavelength range of 400-900 nm. The polyelectrolyte layer was modeled as a Cauchy material with parameters An = 1.61, Bn = 0.01, Cn = 0 such that the modeled refractive index at 630 nm was 1.635.[42, 47] The equivalent thickness for a refractive index-corrected material corresponding more closely to a protein film was modeled from the spectroscopic ellipsometry data by substituting the polyelectrolyte layer in the model for a layer modeled as a Cauchy material with parameters An = 1.47, Bn = 0.01, Cn = 0 corresponding to a refractive index n = 1.495 at 630 nm.[48, 49]

3.3.5 Cell Culture of Aplysia californica Pedal Ganglion Neurons

Sea slug Aplysia californica (100-300 g) were supplied by the National Resource for

Aplysia (Miami, FL) and kept in circulated, aerated seawater at 14°C. Prior to dissection, the animals were anesthetized by injection of isotonic magnesium chloride solution into the body cavity (~30-50% of body weight). Individual Aplysia pedal neurons were isolated after

86 incubation in artificial sea water (ASW; (in mM) 460 NaCl, 10 KCl, 10 CaCl2, 22 MgCl2, 26

MgSO4, and 10 HEPES, pH 7.7) supplemented with proteases (1% type XIV, Sigma-Aldrich) at

34 °C for 60-120 min.[50, 51] Aplysia neurons can be grown on a variety of substrates assuming correct surface chemistry is present and had no difficulties growing on the plasmonic crystal surface.[52-54] A poly-L-lysine layer was formed on the plasmonic crystal surface, and the cells were mechanically isolated in ASW, transferred onto the plasmonic crystal surface immersed in

ASW supplemented with antibiotics (ASW containing 100 units/ml penicillin G, 100 mg/ml streptomycin, and 100 mg/ml gentamicin, pH 7.7) and left to attach and grow overnight at room temperature. Six to eight pedal ganglion neurons were sparsely cultured on the plasmonic crystal surface at a time. Cells were fixed by addition of 1 mL of 4% paraformaldehyde to 3 mL ASW antibiotic culture media and occasional stirring for 30 s, removal of 1 mL of the solution, addition of another 1mL of the 4% paraformaldehyde solution and exposure for 30 s, followed by the removal of all solution. After removal of the fixation solution, the cells were rinsed with deionized water and dried.

3.3.6 Reflection Mode Plasmonic Imaging

Reflection mode images of both the polyelectrolyte films and cells cultured on the plasmonic crystal surface were obtained using an Olympus AX-70 upright microscope with a halogen light source and a 20×, 0.40 NA objective lens. A frosted glass filter was placed immediately in front of the light source to homogenize the incident illumination. The plasmonic crystal was turned upside-down so that light would be incident first on the glass substrate of the plasmonic crystal and then propagate to the nanostructured metal surface. Bandpass filters (500-

550 nm, 525-100 nm, 570-600 nm, 570-1000 nm, and 610-700 nm) were purchased from Omega

Optical and inserted in front of the microscope camera. Images were captured using an Optronics

Magnafire charge-coupled device camera (1280 × 1024 array of pixels, each pixel 6.7 × 6.7 μm).

The halogen lamp intensity was varied to control the overall exposure. Grayscale images were acquired using individual exposure times of 750 ms with fifteen consecutive images averaged together. Image processing and contrast calibration was performed using Matlab and ImageJ.

Vignetting in the reflection images was corrected using images of a silver mirror acquired with each bandpass filter. A Gaussian blur (100 pixel radius) was applied to the reference mirror image to remove blemishes and debris in the image, and the reference image was normalized by dividing each pixel by the averaged pixel intensity in the image.[55] Vignetting in the polyelectrolyte layer and cell images was corrected using a pixel-by-pixel division of the experimental image by the blurred mirror reference image obtained with the corresponding bandpass filter.

3.3.7 Atomic Force Microscopy and Scanning Electron Microscopy of Cells on Plasmonic

Crystals

Atomic force microscopy (AFM) height profiles of Aplysia neurons cultured on the surface of the plasmonic crystal were measured using an Asylum Research MFP-3D atomic force microscope operated in tapping mode. Analysis of the data, including the Fast Fourier Transform filtering, was performed using IgorPro. Scanning electron microscope (SEM) images of Aplysia neurons cultured on plasmonic crystal were acquired using a JEOL 6060-LV scanning electron microscope operated under high vacuum. A thin layer of Au/Pd was sputtered onto the plasmonic crystal and Aplysia neuron after all other analyses were completed to make the sample electrically conductive for SEM imaging.

3.3.8 Finite-Difference Time-Domain (FDTD) Simulations of Plasmonic Nanostructures

3D FDTD simulations were carried out to model the zero-order reflection spectra (with

88 light normally incident on the glass substrate) and their dependence on the polyelectrolyte layer thickness. The unit cell grid spacing was 4 nm in all three dimensions with a total unit cell size of

187×187×600 grid points. The unit cell geometry defined a gold nanohole in the x-y plane with a

740 nm center-to-center hole spacing, 456 nm hole diameter, 292 nm relief depth, 32 nm Au film on the top surface of the plasmonic crystal, 12 nm Au film conformally coating the nanohole sidewalls, and 12 nm Au film on the bottom of the nanoholes. Periodic boundary condition in x and y axis generated an infinite square array, and uniaxial perfectly matched layers were incorporated on the top and bottom surfaces of the unit cell to minimize the effects of unintended reflection from the domain boundaries. The frequency dependent dielectric constant of gold was modeled using previously reported parameters from a Drude plus two-pole Lorentzian model.[36] The refractive indices of NOA, polyelectrolyte, refractive index-corrected material, and air were taken to be 1.56, 1.64, 1.50, and 1.00, respectively.

3.4 Results and Discussion

3.4.1 Plasmonic Crystals and Biological System Model

We examined a model biological system, pedal ganglion neurons isolated from Aplysia californica central nervous system and cultured on a plasmonic crystal. We provide a quantitative analysis of the cell peripheral structural features. Specifically, residual structures associated with the growth cones and cytoplasmic extensions mediating neurite development using contrasts observed in multispectral images. The analytical method described here offers a proof of principle for analytical method that can be applied more broadly in reflection mode imaging studies of complex biological systems. We conducted these studies using a full 3D plasmonic crystal (shown in Figure 3.2) that past work demonstrated possesses molecular

89 sensitivity in imaging mode analyses performed in white light.[46] Work was done on the square array of embossed nanoholes (460 nm diameter, 300 nm depth, 740 nm pitch) conformally coated with 32nm of Au (Figure 3.2a). We begin the discussion by establishing first the means by which plasmonic crystal image contrasts can be measured and quantitatively interpreted within an experimentally validated theoretical model. We follow this by showing an exemplary application of measurements made of plated cells that were fixed after various periods of growth.

3.4.2 Reflection Imaging Contrast Calibration

Polyelectrolyte films comprised of PAH and PSS were assembled on the surface of the plasmonic crystal via layer-by-layer self-assembly.[42, 47] The films were measured to calibrate the reflection contrast changes associated with changes in film thickness. Spectroscopic ellipsometry was used to determine the thickness of dried polyelectrolyte films which were simultaneously deposited onto gold-coated silicon wafer pieces under identical conditions. The results showed that the thickness of the polyelectrolyte assemblies increases by 3.02 ± 0.38 nm for each PAH/PSS deposition step.

The optical response of the plasmonic crystal is influenced by the local refractive index environment and the thickness of the dielectric above the metal surface. For this reason, imaging mode contrast calibrations made using LBL films may complicate quantitative analyses of biological materials due to their markedly different optical properties: the refractive index (n) of the polyelectrolyte is ~ 1.64[42, 47] and that for a biological specimen being more typically is

~1.35-1.5[48, 49]. Ellipsometry, which is sensitive to both layer thickness and refractive index, was used to measure the thicknesses of the polyelectrolyte films and to determine the thickness required for a film with a refractive index more appropriate for biological materials with high organic content (n ~ 1.50) to exhibit similar ellipsometric behavior. For clarity, we operate with

90 a refractive index-corrected thickness for the biological samples (θ), which corresponds to that thickness determined through ellipsometric modeling. Figure 3.3 presents the corresponding pairs for polyelectrolyte and index-corrected thicknesses.

The validity of the ellipsometric thickness reference conversion was evaluated using finite-difference time-domain (FDTD) calculations. Figure 3.4 presents the calculated reflection spectra for a plasmonic crystal coated with a conformal layer of either polyelectrolyte or the equivalent index-corrected material representing the biological material of fixed cells. These data clearly demonstrate the suitability of a single optical data set from the polyelectrolyte LBL assemblies as a calibration reference. The nearly identical reflection spectra show excellent correspondence between the optical properties of the polyelectrolyte and the index-corrected material. The small discrepancies between the spectra likely stem from limitations in the computational model as the equivalent index-corrected thickness was not an exact multiple of the modeled grid spacing of 4 nm.

Reflection images of polyelectrolyte LBL films deposited on the plasmonic crystal surface were acquired with bandpass filters inserted immediately in front of the CCD camera.

The plasmonic crystal was imaged from the reverse side, and a representative image is presented in Figure 3.5a. To correct for variation in illumination intensity between images, a drop of photocurable polyurethane (NOA) was deposited and cured on one corner of plasmonic crystal surface. This NOA-covered region (labeled as NOA in Figure 3.5a) was blocked from subsequent polyelectrolyte depositions, creating a region where the surface refractive index profile remained constant. The region marked LBL in Figure 3.5a denotes the area where the polyelectrolyte was deposited. The dot boxes drawn on the Figure 3.5a illustrate where the average pixel intensity was calculated for the NOA and LBL regions. The normalized pixel

91 values for LBL region in images taken of the each LBL deposition were scaled according to the following relationships. In the first, the average pixel intensity values for the polyelectrolyte-free

NOA regions in all images were equated according to Equation 3.1:

Reference NOA average Scaled LBL average =  Image LBL average (3.1) Image NOA average

The normalized reflection contrast was calculated using the scaled average pixel intensity of the

LBL regions according to Equation 3.2:

[scaled LBL average] - [reference LBL average] Normalized reflection contrast = (3.2) [reference LBL average] where the reference average is defined as the average pixel intensity of the reflection image for the plasmonic crystal before the LBL deposition. Use of the scaled LBL average pixel values corrects for differences in illumination intensity between images, while the normalized reflection contrast expresses the relative reflection contrast change of the LBL region as compared to the reference LBL average pixel intensities. This normalized reflection contrast value may be positive or negative, corresponding to reflectivity higher or lower than the reference condition.

To demonstrate the contrast difference between the polyelectrolyte and polyelectrolyte- free regions and to estimate the lateral resolution of the optical system, Figure 3.5b shows the step-edge profile for the normalized reflection contrast along the yellow arrow shown in Figure

3.5a. A step function was convolved with a Gaussian function and iterated in width to match the observed behavior of the reflection contrast.[41] The resultant fit (red line, as presented in Figure

3.5b) shows that the data can be well modeled with a limiting lateral resolution of 1.0 µm. This value is only slightly larger than the spatial resolution limit of the microscope optics and camera at this magnification, ~0.67 µm.

While the center-to-center spacing between holes is ~740 nm, depending on the optical

92 alignment, a single pixel in the image of the plasmonic crystal may overlap part of one or several nanoholes. It is reasonable that the reflectivity may not be identical between the top and bottom surfaces of the nanohole, and the pixel intensity could vary somewhat due to differences in nanohole coverage. All the same, we noted no particular patterns appearing in histograms of the pixel intensities in the calibration images.

The normalized reflection contrast (NRC) change as a function of refractive index- corrected thickness is presented in Figure 3.6. The reflection contrast becomes more negative as the surface layer thickness (θ) increases when a 500-550 nm bandpass filter is used, while the reflection contrast becomes more positive when imaged using a 525-1000 nm bandpass filter.

These experimental data demonstrate a ‘contrast inversion’ in which the reflection intensity changes increase or decrease depending on the wavelengths used. Because a simple linear regression is more adequate to mathematically interpret this experimental behavior, linear regressions were calculated for both wavelength ranges: NRC = − 0.00224θ + 0.03183 for the

500-550 nm bandpass filter (black line in Figure 3.6), and NRC = 0.00085θ + 0.01872 for the

525-1000 nm bandpass filter (red line in Figure 3.6). More complex reflection contrast behaviors were observed at other wavelength ranges. Using a 570-1000 nm bandpass filter, for example, a non-monotonic dependence is seen in which the reflection contrast initially increases as the surface layer thickness increases but then reverses and begins to decrease as the coverage was further incremented. This complexity is attributed to that of the underlying surface plasmon modes and their changes in response to changes in the refractive index profile of the adsorbed material on the plasmonic crystal surface.

Additional FDTD calculations were performed to determine changes in the optical responses of the plasmonic crystal with increasing polyelectrolyte thickness. Figure 3.7a presents

93 the theoretically calculated reflection spectra for a plasmonic crystal of the experimental geometry as the thickness of polyelectrolyte was changed. These spectra demonstrate a complex behavior akin to that seen in the reflection imaging contrast calibrations. Some wavelength ranges (e.g. ~675-770 nm) exhibit a decrease in reflectance with thicker polyelectrolyte films while the opposite trend is observed at other wavelength ranges (e.g. ~770-820 nm). These trends are similar to those seen in experiment.

For comparison, the simulated reflectance spectra for polyelectrolyte films on flat gold surfaces are presented in Figure 3.7b. Above ~530 nm, the reflectance intensity decreases nearly monotonically as the polyelectrolyte layers thickness increases, although the magnitudes of the reflectance changes do exhibit a dependence on wavelength. This flat film modeling accounts for

Fresnel effects associated with light interacting with and passing through boundaries between layers with different refractive indices but otherwise lack plasmonic properties. The absence of a wavelength dependent ‘contrast inversion’ for this flat film akin to that observed experimentally points towards surface plasmon resonances as the origin of the latter effects.

It is important to note that the results presented in Figure 3.7 were obtained from simulations in which the illumination was normally incident to the surface. The reflection images obtained experimentally used an objective lens with a numerical aperture of 0.40, corresponding to a cone of illumination with a maximum angle of ~23°. This is significant because the spectral properties of the plasmonic crystal are known to depend on illumination angle.[40] While the computational modeling performed here offers insight into the complexity behind the observed reflection behavior, a more complete theoretical understanding will require simulations that incorporate those oblique illumination angles, which is not currently possible within our FDTD code.

3.4.3 Reflection Imaging of Fixed Aplysia Pedal Ganglion Neurons on Plasmonic Crystals

To explore the quantitative features of SPR-based reflection imaging of a complex biological specimen, Aplysia pedal ganglion neurons were cultured on the surface of a plasmonic crystal (shown in Figure 3.8). Figures 3.8a and Figure 3.8b present reflection images of a representative fixed neuron cultured on a nanostructured plasmonic crystal surface, while Figure

3.9a and Figure 3.9b present images of a cell cultured on a flat gold surface. A visual comparison of the images collected using a 500-550 nm bandpass filter (Figure 3.8a and Figure 3.9a) and a

525-1000 nm bandpass filter (Figure 3.8b and Figure 3.9b) clearly reveals differences in contrast, including its inversion, in a manner predicted by the computational modeling. The peripheral regions of the fixed neurons on flat gold appear darker than the surrounding area in both the 500-

550 nm and 525-1000 nm images, while the analogous regions of the cell grown on the plasmonic crystal show an inversion in contrast where the peripheral region appears darker than the surrounding area in the 500-550 nm image but lighter than the surrounding area in the 525-

1000 nm image.

These fixed neurons have features that range in thickness from tens of nanometers to several microns. As surface plasmon effects are generally sensitive to refractive index profile changes only within ~100-200 nm from the metal surface, many of the features seen in the images of Figure 3.8 must have other origins. The neuron’s body (especially the several tens of microns thick regions) on the plasmonic crystal appears dark and possesses circumferential bright fringes in the reflection images. These are likely the result of a convolution of various optical phenomena, including: light absorption, reflection, and scattering by cellular structures; surface plasmon-mediated reflection; and interference effects. To highlight the plasmonic resonance phenomena manifested in the images of Figure 3.8, we focus only on those features

95 whose dimensions are within the range where the linear regression of optical response as a function of thickness is valid (i.e. ~80nm from the surface).[42]

The calibration between reflection contrast and index-corrected thickness presented in

Figure 3.6 were applied to the reflection images shown in Figure 3.8a and Figure 3.8b to determine the thicknesses of the regions of the fixed Aplysia pedal neurons behaving as thin films. The normalized reflection contrast was calculated pixel-by-pixel in the cell image using an empty area with no apparent features immediately adjacent to the cell as the reference reflection.

The normalized reflection contrast was then transformed mathematically to an index-corrected material thicknesses value using the appropriate regression equation corresponding to the bandpass filter used to acquire the original image. These thickness-transformed cell images are presented in Figure 3.8c (for the 500-550 nm bandpass filter) and Figure 3.8d (for the 525-1000 nm bandpass filter).

In calculating the thickness transformations, the normalized reflection contrast values were restricted to 휃 lying between 0 and 80 nm. Pixel values outside of this range were truncated, resulting in regions of the image being saturated at either the minimum or maximum reflection contrast value. Because the soma has a thickness significantly larger than the plasmonic sensing volume, the thickness estimates of these regions are clearly invalid. Therefore, the analysis performed in this work focuses instead on the thinner neurite outgrowths and growth cone regions which are arguably more interesting in the study of cell growth on the substrate.

Images from both wavelength regions reveal a sparse film of material in this region with a thickness of ~30 nm and string-like filaments with thicknesses of ~60 nm. We describe origins for these features in sections that follow.

We note that ‘empty’ regions in the image (with no apparent cell structures) actually have

96 a thin film of poly-L-lysine, which had been applied to the plasmonic crystal surface to improve the biocompatibility of the substrate. The contributions of this poly-L-lysine layer to the estimated thickness can be neglected, however, because its thickness is accounted for when calculating the normalized reflection contrast.

The reliability of the thickness evaluation protocol can be evaluated via the self- consistency of the estimates of film coverage obtained using different bandpass filters. Figures

3.8e and 3.8f demonstrate identical thickness profiles (휃) along the white arrows on cell images obtained using the 500-550 nm and 525-1000 nm bandpass filters, respectively. The effect of truncating the contrast values in the image prior to applying thickness transformation is particularly apparent in the line profile for the 525-1000 nm image (in Figure 3.8f) where many pixels were assigned a thickness of zero. Despite the truncation in the 525-1000 nm image, good agreement is observed for the height of the filament feature at position X ~ 40 μm which both wavelength ranges assign a thickness of ~60 nm. Good correspondence is also observed for the feature at X ~ 110 μm where the surface feature thickness is approximately ~50 nm. However, the agreement between the two images is poorer for features with smaller values; for example, small features were not observed in the 500-550 nm image as in the 525-1000 nm image between

X ~ 50-90 μm. This discrepancy may result from the relatively coarse linear regressions applied to the calibration imaging data, which may underestimate or overestimate the contrast change in particular thickness ranges. Using a more finely-grained regression analysis or more complex mathematical fits would be expected to yield improved consistency between wavelength regions.

Alternatively, as discussed in sections that follow, multispectral image contrasts can be calculated from the component bandpass filter data that provide greatly enhanced capabilities for quantitative imaging.

3.4.4 Atomic Force Microscopy (AFM) Height Profiles

In addition to comparing profiles measured using different bandpass filters, atomic force microscopy (AFM) imaging was performed as an independent verification of the height profiles of the fixed neuron cells structures calculated using data from reflection plasmonic imaging.

Representative AFM data are presented in Figure 3.10a, with a three-dimensional representation of the AFM data shown on Figure 3.10b. These images reveal a periodic modulation which arises from the underlying nanohole array that comprises the plasmonic crystal. The periodic modulation from the underlying substrate is not observed in the reflection images because the individual nanoholes (~460 nm in diameter) are smaller than the length scales resolved by the camera at this magnification (~670 nm) and the likely ‘smearing’ due to the comparable length scale of the LSPRs that the plasmonic crystal supports.[28, 29]

The periodic modulation in the AFM images resulting from the plasmonic crystal topography adversely affects the analysis of the thicknesses of cellular features on the surface.

To overcome this, the periodic oscillations were filtered out using a Fast Fourier Transform

(FFT) of the image where all but the lowest frequency components were removed. A 3D projection of the FFT filtered data is presented in Figure 3.10c. There is a loss resolution that occurs as a result of the FFT filtering. A scanning electron micrograph of a similar fixed cell is presented in Figure 3.10d. This image shows that the thinner cell outgrowth regions conform, at least partially, to the plasmonic crystal’s surface topography while thicker layers appear to be less strongly affected, structured features similar to those seen in the AFM data.

A comparison of height profiles is presented in Figure 3.10f for the AFM image and in

Figure 3.10g for the plasmonic reflection image. These profiles were taken from the same cell area represented by the white arrows superimposed on the AFM image in Figure 3.10a and the

98 reflection image in Figure 3.10e. Comparing the two height profiles reveals a marked correspondence in the thickness estimates of the filament structures at X ~ 30 µm in the AFM profile (Figure 3.10f) and at X ~ 30 µm in the reflection image profile (Figure 3.10g); both methods assign the feature a thickness of ~60 nm. These AFM data, thus provide independent confirmation of the thicknesses estimated using the plasmonic reflection contrast calibrations and of the estimated refractive index of the fixed cell material. The latter value, being higher than the refractive index of live cell components (where n ~ 1.35 -1.41), reflects the loss of water from the materials present in the fixed cell.

3.4.5 Increased Image Contrast through Wavelength Combination

A charge-coupled device (CCD) camera was used as the imaging detector in this work, which integrates the photon flux across the accessible wavelength range to report pixel intensity.

Reflectivity changes in the sample result in changes in the photon flux and are interpreted as reflectance changes in the image, and an increase in the number of photons at one wavelength within that accessible range of reflected light can be offset by a decrease in photons at a different wavelength within that range. The integrative nature of the CCD camera can ultimately result in reduced image contrasts if the wavelengths used overlap spectral regions with opposite contrast behavior (where the reflectance increases at one wavelength but decreases at another). Bandpass filters allow these spectral regions to be isolated from one another and analyzed separately in a form of quasi-hyperspectral imaging, albeit one with relatively coarse wavelength control. By identifying wavelength regions where the reflection contrast exhibits opposite behaviors and imaging these regions separately, recombination of the data from those wavelength regions can be used to restore or even enhance the contrast.

Figure 3.11 shows reflection contrast calibration curves obtained using 570-600 nm (pink

99 square) and 570-1000 nm (yellow-green triangle) bandpass filters. Due to the wavelength overlap between the two filters, the reflection behavior between 570-600 nm is included (and convolved within) the optical data collected over the 570-1000 nm wavelength range. These calibration curves indicate that the reflection contrast decreases at θ greater than ~20 nm in both wavelength ranges, but the contrast decreases obtained using the 570-1000 nm filter are not as pronounced as those found in the 570-600 nm wavelength range. This indicates that the reflection must tend to increase for thicker θ at wavelengths between 600-1000 nm in order to offset the contrast decreases that occur over the 570-600 nm range. Most interestingly, the 570-

1000 nm data show a larger increase in reflection contrast for thinner θ (up to ~20 nm) than is seen in the 570-600 nm data. It can be concluded as a result that the contrast scaling seen over the 570-600 nm and 600-1000 nm ranges are, in fact, opposite in sign. This provides a means to enhance the overall imaging contrast by recombining images captured over these different wavelength ranges.

The wide spectral response of the camera is needed in order to properly combine the calibration images obtained using different bandpass filters into a single composite calibration curve. The spectral response of the CCD detector is not equal across all wavelengths; its sensitivity peaks between 500-520 nm and then decreases with increasing wavelength of the light

(as shown in Figure 3.12).[56] The 570-600 nm bandpass filter covers only ~7% of the wavelength range between 570-1000 nm but accounts for ~18% of the total sensitivity. Thus it results that the contributions of the 570-600 nm wavelength range to the overall image contrast are disproportionately larger relative to the size of the bandpass wavelength range. With consideration of the spectral sensitivity of the CCD camera, the individual images made using the 570-600 nm and 570-1000 nm bandpass filters were weighted by the total integrated

100 sensitivity of camera over the specific wavelength range and combined according to Equation

3.3:

([570-1000 nm] 147.75)  2  ([570-600 nm]  26.55) Composite Image  (3.3) 147.75

The value of 147.75 originates from the total integrated sensitivity from 570-1000 nm (green + pink area) presented on the spectral sensitivity plot in Figure 3.12, while the 26.55 value is the integrated sensitivity from 570-600 nm (green area) in Figure 3.12. The image data from the 570-

600 nm wavelength range was subtracted twice in this analysis (via a factor of two in the formula above) – the first subtraction removes its contribution from the data collected over the entire

570-1000 nm wavelength range, and the second subtraction restores its contribution but with an inverted sign. The resultant image was then used to calculate the normalized reflection contrast as described previously, and the combined contrast calibration results are plotted as the dark cyan circles in Figure 3.11. It is readily apparent that the overall reflection contrast increases across the range of index-corrected material thicknesses (θ) examined. A power law regression was applied to model the combined calibration curve, with a best fit of NRC = 0.01565θ0.6323.

This composite calibration curve was applied to images of the fixed Aplysia neurons on the plasmonic crystal (which had been combined in a similar manner), and the resultant thicknesses estimates are presented in Figure 3.13a. The thickness profile in Figure 3.13b (along the white arrow shown in Figure 3.13a) shows good agreement with those produced using the

500-550 nm and 525-1000 nm calibration curves (in Figure 3.8e and Figure 3.8f, respectively).

The smaller features lying between X ~50-90 μm, which were not observed in the 500-550 nm image, are clearly resolved in the combination reflection image. These results show the efficacy of combining images acquired in a multispectral form to increase contrast and sensitivity in the resultant image.

101

3.4.6 Quantitative Thickness Estimates from Non-plasmonic Optical Effects

While the thickness calibration methods demonstrated here rely on the sensitivity of surface plasmon resonances to changes in the refractive index profile at and near the metal surface, the reflection images also contain other areas of the cell with thicknesses significantly greater than the range of the surface plasmons’ evanescent electric field. Additional and quantitatively useful non-plasmonic optical effects are present in these regions of the image.

These include Fresnel reflections from interfaces between materials with different refractive indices along with light scattering and absorption by cellular components. The region boxed in red on the cell image presented in Figure 3.14a (acquired using a 500-550 nm bandpass filter) highlights a notable non-plasmonic optical effect seen in the reflection image. Rotated and magnified views of this region acquired using 500-550 nm, 570-600 nm, and 610-700 nm bandpass filters are presented in Figure 3.14b, respectively. These images reveal alternating light and dark bands moving from the lower left tip towards the upper right. The size and position of these bands changes depending on the particular bandpass filter used.

These alternating intensity bands represent a thin film interference effect (referred to as

Newton’s rings) arising from constructive and destructive interference from light reflected within a resonant cavity formed by the top and the bottom surfaces of the cell structure. Constructive interference of light rays reflected from both surfaces produces bright fringes, while dark fringes appear where destructive interference occurs. The number and position of these bright and dark regions are functions of the wavelengths of light being imaged and the cavity length (cell thickness) required to produce the resonant conditions. Since the wavelengths of light being imaged are known, the thin film interferences provide an additional method to quantitatively estimate the thicknesses of thin film structures presented in the sample. Figure 3.15 illustrates the

102 path length difference originating from reflections of light propagating through a medium with

refractive index n1 and reflecting from two surfaces – the top and bottom surfaces of a thin film

with a thickness d and a refractive index n2 . The optical path difference (OPD) is given by the expression below:

OPD n21 AB  BC  n AD (3.4)

d where AB BC and AD 2 n2 d tan 2 sin 1 . cos2

The OPD equation can be simplified using Snell’s law ( nn1sin 1 2 sin 2 ) to

OPD 2 n22 d cos (3.5)

At the bright fringe, the optical path difference must be an integer multiple of the wavelength

(mλ, where m is an integer).

m 2 n22 d cos (3.6)

m d  (3.7) 2n22 cos

Figure 3.14a was acquired using a 500-550 nm bandpass filter and the incident angle of light (θ1) ranges from 0° to 23.6° due to the numerical aperture of the objective lens (NA = 0.40).

Therefore, the thickness of the cellular film ( = 1.50) at the bright fringes can be roughly estimated and shown in Table 3.1.

Table 3.1 The thickness of the cellular film ( = 1.50) at the bright fringes in 500-550 nm image.

λ = 500 nm λ = 550 nm 0° 23.6° 0° 23.6° m=1 167 nm 173 nm 183 nm 190 nm Thickness of protein m=2 333 nm 346 nm 366 nm 380 nm m=3 500 nm 519 nm 549 nm 571 nm

103

The resonant cavity length was estimated for the second band indicated in the 500-550 nm image (in Figure 3.14b) using the extreme ends of the range of wavelengths and incident angles (0° and 23.6°) along with the previous estimate of the refractive index (n ~ 1.5). The convolution and overlap of different wavelengths with different illumination angles predict a range of cavity thicknesses lying between ~300-400 nm as being capable of producing constructive interference at the location highlighted in 500-550nm image (in Figure 3.14b).

AFM measurements were made to independently evaluate the latter thickness estimate. A representative FFT filtered AFM image is presented in Figure 3.16a and a height profile along the blue arrow in Figure 3.16a is shown in Figure 3.16b. The thickness revealed by AFM data is between 300 nm and 400nm, in good agreement with the thickness analysis for the data in Figure

3.14b. This interferometric thickness calculation is a simple complement to the surface plasmon- based measurements, capable of measuring surface films thicker than the plasmonic sensing volume without prior calibration. However, the thickness range estimates can become very broad when applied to images acquired using large wavelength ranges, and the nature of the microscope objectives themselves results in a range of thicknesses capable of producing constructive or destructive interference at a location (even under monochromatic illumination).

The minimum thickness required to observe thin film interference effects are one-half of the wavelength (or one-fourth if reflection from the upper surface results in a 180° phase shift). Even for 400 nm illumination at normal incidence, this corresponds to a minimum thickness of 100 nm

– coincidentally the upper limit for quantitative thickness measurements using plasmonic calibrations. While these considerations constrain the use of thin film interference effects to quantitate material thicknesses, they also suggest possibilities for a synergistic integration of both methods of analysis.

104

3.4.7 Inferences Regarding Cell Structures Observed on Plasmonic Crystals

The growth and extension of neuritis from the Aplysia pedal ganglion neuron body begins with the emergence of growth cones.[57] The growth cone is defined by a thicker central region, where vesicles and microtubules can be found, and a surrounding thinner ‘veil’ composed of lamellipodia and filopodia.[57-59] As the cell matures, materials in the peripheral regions are resorbed as the neurite thickens and rounds,[57, 60] and it is likely that quantitative analysis of reflection images of neuronal cell structures at different developmental stages might help to elicit new understandings of the neurite outgrowth process.

The images presented in Figure 3.17 illustrate this notional point, showing fixed Aplysia pedal ganglion neurons having regions at presumably different stages of growth. The structure labeled ‘A’ in Figure 3.17a (thickness calibrated reflection image using combination of 570-600 nm and 570-1000 nm images) appears to be a neurite which has fully formed. The area labeled

‘B’ appears to be a sparse distribution of thin features and may be the remnants of the peripheral region of the growth cone. Similarly, the darker neurite on the left side (labeled ‘C’) is readily distinguished from the thinner region (labeled ‘D’), and the patchy nature of the surface coverage moving further away from the cell body (labeled ‘E’) may be indicative of the resorption or partial resorption of the peripheral portion of the growth cone (in Figure 3.17b). The cell structures characterized by ‘Newton’s ring’ in the thin film interference analysis may be the central regions of the growth cones (red asterisks in Figure 3.17c), which are surrounded by the thin ‘veil’ previously associated with the peripheral region of the growth cone. The neurons in

Figures 3.17a and 3.17b have more clearly defined neurites and fewer central regions and thus may be more mature, while the less developed neuron shown in Figure 3.17c has a morphology characterized by fewer defined neurites and many more growth cone regions.

105

3.5 Conclusions

We describe a label-free imaging technique that utilizes a plasmonic crystal in conjunction with bandpass filters to quantify the thickness of thin film structures present in a model complex biological structure, pedal ganglion neurons from Aplysia californica that were fixed after growth in culture. Contrast calibrations using polyelectrolyte layer-by-layer assemblies were applied to estimate the thickness of thin growth regions of the neurons on the plasmonic crystals. The reliability of the imaging method was confirmed with height profiles acquired by atomic force microscopy. The use of bandpass filters to restrict the wavelengths of light imaged can improve image contrast and sensitivity, with further improvements possible by a combination of images acquired with different bandpass filters. Computational studies provide insight into contrast changes associated with plasmonic phenomena as a function of film thickness. Films having thicknesses beyond the limit of this SPR method can be analyzed using alternative non-plasmonic optical effects, such as Fresnel reflection, scattering, and absorption.

Further improvements to the image contrast and sensitivity may be achieved through optimization of the plasmonic crystal structure. This plasmonic reflection imaging method has the potential to be adapted for live cell imaging, because it does not demand the specific operating conditions and long scanning times. While this imaging technique was utilized to analyze the thickness of a fixed cell in this study, this quantitative imaging method shows considerable promise as a label-free analytical technique for studying cell morphology or cell dynamics stimulated by chemical or physical changes in real time. Studies are in progress to explore these applications.

106

3.6 Acknowledgements

The authors thank Xiying Wang for preparing the Aplysia cell cultures. This work was supported by the U.S. Department of Energy under prime Award No. DE-SC0001293 via subcontract 67N-1087758 from California Institute of Technology as part of the DOE Energy

Frontier Research Center on Light-Material Interactions in Energy Conversion and was performed using resources at the Frederick Seitz Materials Research Laboratory Central

Facilities at the University of Illinois, including the Center for Microanalysis of Materials; the

Micro/Nanofabrication Facility; and the Laser and Spectroscopy Facility. We also acknowledge the use of the Turing cluster, maintained and operated by the Computational Science and

Engineering Program at the University of Illinois.

107

3.7 References

1. Barry, R.A., et al., Direct-Write Assembly of 3D Hydrogel Scaffolds for Guided Cell Growth. Advanced Materials, 2009. 21(23): p. 2407-2410. 2. Millet, L.J., et al., Microfluidic devices for culturing primary mammalian neurons at low densities. Lab Chip, 2007. 7(8): p. 987-994. 3. Millet, L.J., et al., Guiding neuron development with planar surface gradients of substrate cues deposited using microfluidic devices. Lab Chip, 2010. 10(12): p. 1525- 1535. 4. Zhang, H.B., J.N.H. Shepherd, and R.G. Nuzzo, Microfluidic contact printing: a versatile printing platform for patterning biomolecules on hydrogel substrates. Soft Matter, 2010. 6(10): p. 2238-2245. 5. Shepherd, J.N.H., et al., 3D Microperiodic Hydrogel Scaffolds for Robust Neuronal Cultures. Advanced Functional Materials, 2011. 21(1): p. 47-54. 6. Corum, L.E., et al., Using Microcontact Printing of Fibrinogen to Control Surface- Induced Platelet Adhesion and Activation. Langmuir, 2011. 27(13): p. 8316-8322. 7. Hanson, J.N., et al., Textural guidance cues for controlling process outgrowth of mammalian neurons. Lab Chip, 2009. 9(1): p. 122-131. 8. Conn, P.M., Imaging and spectroscopic Analysis of Living Cells: Optical and Spectroscopic Techniques. Vol. 504. 2012: Elsevier Science: New York. 1-506. 9. Barty, A., et al., Quantitative optical phase microscopy. Opt Lett, 1998. 23(11): p. 817- 819. 10. Allen, R.D., N.S. Allen, and J.L. Travis, Video-Enhanced Contrast, Differential Interference Contrast (Avec-Dic) Microscopy - a New Method Capable of Analyzing Microtubule-Related Motility in the Reticulopodial Network of Allogromia Laticollaris. Cell Motility and the Cytoskeleton, 1981. 1(3): p. 291-302. 11. Dodt, H.U. and W. Zieglgansberger, Visualizing Unstained Neurons in Living Brain- Slices by Infrared Dic-Videomicroscopy. Brain Res, 1990. 537(1-2): p. 333-336. 12. Pluta, M., Advanced light microscopy1988: Elsevier: New York. 13. Curtis, A.S., Cell reactions with biomaterials: the microscopies. Eur Cell Mater, 2001. 1: p. 59-65. 14. Pauletti, G., et al., Assessment of methods for tissue-based detection of the HER-2/neu alteration in human breast cancer: A direct comparison of fluorescence in situ hybridization and immunohistochemistry. Journal of Clinical Oncology, 2000. 18(21): p. 3651-3664. 15. Giepmans, B.N.G., et al., Review - The fluorescent toolbox for assessing protein location and function. Science, 2006. 312(5771): p. 217-224. 16. Hu, C.D. and T.K. Kerppola, Simultaneous visualization of multiple protein interactions in living cells using multicolor fluorescence complementation analysis. Nature Biotechnology, 2003. 21(5): p. 539-545. 17. Zhang, S.F., C. Ma, and M. Chalfie, Combinatorial marking of cells and organelles with reconstituted fluorescent proteins. Cell, 2004. 119(1): p. 137-144. 18. Stephens, D.J. and V.J. Allan, Light microscopy techniques for live cell Imaging. Science, 2003. 300(5616): p. 82-86. 19. Tanaami, T., et al., High-speed 1-frame/ms scanning confocal microscope with a microlens and Nipkow disks. Appl Opt, 2002. 41(22): p. 4704-4708. 108

20. Nakano, A., Spinning-disk confocal microscopy - A cutting-edge tool for imaging of membrane traffic. Cell Structure and Function, 2002. 27(5): p. 349-355. 21. Semwogerere, D. and E.R. Weeks, Confocal Microscopy, in Encyclopedia of biomaterials and biomedical engineering, G.E. Wnek and G.L. Bowlin, Editors. 2008, Informa Healthcare USA: New York. 22. Marquet, P., et al., Digital holographic microscopy: a noninvasive contrast imaging technique allowing quantitative visualization of living cells with subwavelength axial accuracy. Opt Lett, 2005. 30(5): p. 468-470. 23. Wang, Z., et al., Spatial light interference microscopy (SLIM). Opt Express, 2011. 19(2): p. 1016-1026. 24. Rotsch, C. and M. Radmacher, Drug-induced changes of cytoskeletal structure and mechanics in fibroblasts: An atomic force microscopy study. Biophys J, 2000. 78(1): p. 520-535. 25. Radmacher, M., et al., Measuring the viscoelastic properties of human platelets with the atomic force microscope. Biophys J, 1996. 70(1): p. 556-567. 26. Curtis, M.W. and B. Russell, Micromechanical regulation in cardiac myocytes and fibroblasts: implications for tissue remodeling. Pflugers Archiv-European Journal of Physiology, 2011. 462(1): p. 105-117. 27. Homola, J., Surface plasmon resonance sensors for detection of chemical and biological species. Chem Rev, 2008. 108(2): p. 462-493. 28. Stewart, M.E., et al., Nanostructured plasmonic sensors. Chem Rev, 2008. 108(2): p. 494-521. 29. Stewart, M.E., et al., Quantitative multispectral biosensing and 1D imaging using quasi- 3D plasmonic crystals. Proc Natl Acad Sci U S A, 2006. 103(46): p. 17143-17148. 30. Jung, L.S., et al., Quantitative interpretation of the response of surface plasmon resonance sensors to adsorbed films. Langmuir, 1998. 14(19): p. 5636-5648. 31. Homola, J., S.S. Yee, and G. Gauglitz, Surface plasmon resonance sensors: review. Sensors and Actuators B-Chemical, 1999. 54(1-2): p. 3-15. 32. Johnsson, B., S. Lofas, and G. Lindquist, Immobilization of Proteins to a Carboxymethyldextran-Modified Gold Surface for Biospecific Interaction Analysis in Surface-Plasmon Resonance Sensors. Anal Biochem, 1991. 198(2): p. 268-277. 33. Jonsson, U., et al., Real-Time Biospecific Interaction Analysis Using Surface-Plasmon Resonance and a Sensor Chip Technology. Biotechniques, 1991. 11(5): p. 620-627. 34. Anker, J.N., et al., Biosensing with plasmonic nanosensors. Nature Materials, 2008. 7(6): p. 442-453. 35. Jordan, C.E., et al., Surface plasmon resonance imaging measurements of DNA hybridization adsorption and streptavidin/DNA multilayer formation at chemically modified gold surfaces. Anal Chem, 1997. 69(24): p. 4939-4947. 36. Aldred, N., et al., Real-Time Quantification of Microscale Bioadhesion Events In situ Using Imaging Surface Plasmon Resonance (iSPR). ACS Appl Mater Interfaces, 2011. 3(6): p. 2085-2091. 37. Moh, K.J., et al., Surface plasmon resonance imaging of cell-substrate contacts with radially polarized beams. Opt Express, 2008. 16(25): p. 20734-20741. 38. Wang, W., et al., Mapping Single-Cell-Substrate Interactions by Surface Plasmon Resonance Microscopy. Langmuir, 2012. 28(37): p. 13373-13379. 39. Lee, H.J., D. Nedelkov, and R.M. Corn, Surface plasmon resonance imaging

109

measurements of antibody arrays for the multiplexed detection of low molecular weight protein biomarkers. Anal Chem, 2006. 78(18): p. 6504-6510. 40. Malyarchuk, V., et al., Spatially resolved biosensing with a molded plasmonic crystal. Applied Physics Letters, 2007. 90(20): p. 203113−203115. 41. Yao, J.M., et al., Seeing molecules by eye: Surface plasmon resonance imaging at visible wavelengths with high spatial resolution and submonolayer sensitivity. Angewandte Chemie-International Edition, 2008. 47(27): p. 5013-5017. 42. Stewart, M.E., et al., Multispectral thin film biosensing and quantitative imaging using 3D plasmonic crystals. Anal Chem, 2009. 81(15): p. 5980-5989. 43. Truong, T.T., et al., Nanopost plasmonic crystals. Nanotechnology, 2009. 20(43): p. 434011. 44. Maria, J., et al., Optimization of 3D Plasmonic Crystal Structures for Refractive Index Sensing. Journal of Physical Chemistry C, 2009. 113(24): p. 10493-10499. 45. Yao, J., et al., Soft embossing of nanoscale optical and plasmonic structures in glass. ACS Nano, 2011. 5(7): p. 5763-5774. 46. Yao, J., et al., Functional nanostructured plasmonic materials. Adv Mater, 2010. 22(10): p. 1102-1110. 47. Caruso, F., et al., Ultrathin multilayer polyelectrolyte films on gold: Construction and thickness determination .1. Langmuir, 1997. 13(13): p. 3422-3426. 48. Voros, J., The density and refractive index of adsorbing protein layers. Biophys J, 2004. 87(1): p. 553-561. 49. Benesch, J., A. Askendal, and P. Tengvall, The determination of thickness and surface mass density of mesothick immunoprecipitate layers by null ellipsometry and protein (125)iodine labeling. Journal of Colloid and Interface Science, 2002. 249(1): p. 84-90. 50. Ye, X.Y., et al., Production of Nitric Oxide within the Aplysia californica Nervous System. Acs Chemical Neuroscience, 2010. 1(3): p. 182-193. 51. Wang, L.P., et al., A Novel Pyridoxal 5 '-Phosphate-dependent Amino Acid Racemase in the Aplysia californica Central Nervous System. Journal of Biological Chemistry, 2011. 286(15): p. 13765-13774. 52. Romanova, E.V., et al., Engineering the morphology and electrophysiological parameters of cultured neurons by microfluidic surface patterning. Faseb Journal, 2004. 18(9): p. 1267-1269. 53. Romanova, E.V., et al., Self-assembled monolayers of alkanethiols on gold modulate electrophysiological parameters and cellular morphology of cultured neurons. Biomaterials, 2006. 27(8): p. 1665-1669. 54. Jo, K., et al., Mass spectrometric imaging of peptide release from neuronal cells within microfluidic devices. Lab Chip, 2007. 7(11): p. 1454-1460. 55. Leong, F.J.W.M., M. Brady, and J.O. McGee, Correction of uneven illumination (vignetting) in digital microscopy images. Journal of Clinical Pathology, 2003. 56(8): p. 619-621. 56. Sony ICX085AL 2/3-in. Progressive Scan CCD Image Sensor with Square Pixel for B/W Cameras, Datasheet No. E95Z10C73. Available from: http://www.alldatasheet.com/datasheet-pdf/pdf/47409/SONY/ICX085AL.html. 57. Lovell, P. and L.L. Moroz, The largest growth cones in the animal kingdom: an illustrated guide to the dynamics of Aplysia neuronal growth in cell culture. Integrative and Comparative Biology, 2006. 46(6): p. 847-870.

110

58. Bamburg, J.R., Introduction to cytoskeletal dynamics and pathfinding of neuronal growth cones. Journal of Histochemistry & Cytochemistry, 2003. 51(4): p. 407-409. 59. Burmeister, D.W. and D.J. Goldberg, Micropruning - the Mechanism of Turning of Aplysia Growth Cones at Substrate Borders Invitro. Journal of Neuroscience, 1988. 8(9): p. 3151-3159. 60. Kedem, O., A. Vaskevich, and I. Rubinstein, Improved Sensitivity of Localized Surface Plasmon Resonance Transducers Using Reflection Measurements. Journal of Physical Chemistry Letters, 2011. 2(10): p. 1223-1226.

111

3.8 Figures

Figure 3.1 Schemes of the soft nanoimprint lithography process: (i) A nanostructured PDMS stamp is cast from a square nanohole array SOG master and used to imprint into an NOA prepolymer on a glass slide. (ii) The stamp is pressed into the NOA prepolymer and cured under ultraviolet light. (iii) The stamp is removed, leaving a replica of the square nanohole array from the original substrate. (iv) 32 nm of gold is sputtered onto the patterned NOA surface to create the plasmonic crystal.

112

Figure 3.2 (a) The SEM image of the plasmonic crystal with diameters and periodicities of ~460 and 740 nm. The scale bar corresponds to 5 μm. Inset: Tilted SEM image of red box. The scale bar corresponds to 100 nm. (b) Top: 3D schematic of the plasmonic crystal. Bottom: 2D schematic of cross section of plasmonic crystal.

113

Figure 3.3 Corresponding pairs of thicknesses for polyelectrolyte films and index-corrected material (n = 1.5) (θ) determined via ellipsometry.

114

Figure 3.4 Finite-difference time-domain (FDTD) simulated reflection spectra for pairs of optically equivalent polyelectrolyte and index-corrected material for the fixed cell conformally covering a 30 nm of gold coated plasmonic crystal: (a) 20 nm polyelectrolyte and 24 nm index- corrected material; (b) 24 nm polyelectrolyte and 28 nm index-corrected material; (c) 40 nm polyelectrolyte and 48 nm index-corrected material.

115

Figure 3.5 (a) Reflection image (using a 500-550 nm bandpass filter) of a polyelectrolyte layer- by-layer assembly (LBL) and photocured polyurethane drop (NOA) on a plasmonic crystal used for the reflection contrast calibration. Average pixel intensities were calculated in the regions marked with red box (NOA region) and blue box (LBL region), and used to determine the reflection contrast calibration. The scale bar corresponds to 100 μm. (b) Step-edge profile of normalized reflection contrast (blue circles) along the yellow arrow in Figure 3.5a and fitted red curve with a Gaussian width of 1.0 μm.

116

Figure 3.6 Normalized reflection contrast (NRC) as a function of index-corrected thickness (θ) using three different bandpass filters: 500-550 nm (dark cyan circles), 525-1000 nm (pink squares), and 570-1000 nm (yellow-green triangles). The black line shows the best fit linear regression for the 500-550 nm bandpass filter (NRC = −0.00224θ + 0.03183) and red line shows that for the 525-1000 nm bandpass filter (NRC = 0.00085θ + 0.01872). The error bars are obscured by the data markers themselves as a result of the large number of pixels (more than 50,000) used to determine the value.

117

Figure 3.7 (a) FDTD calculated reflection spectra for varying thicknesses of polyelectrolyte assemblies on a nanostructured plasmonic crystal. (b) FDTD calculated reflection spectra for varying thicknesses of polyelectrolyte assemblies on flat gold film.

118

Figure 3.8 (a, b) Reflection images of an Aplysia neuron cultured on a plasmonic crystal surface acquired using (a) 500-550 nm bandpass filter and (b) 525-1000 nm bandpass filter. The scale bar in (a) and (b) corresponds to 100 μm. (c, d) Index-corrected thickness (θ) calibrations applied to reflection images of an Aplysia neuron cultured on a plasmonic crystal acquired using (c) 500- 550 nm bandpass filter and (d) 525-1000 nm bandpass filter. The z-scale for image is restricted to the plasmonic sensing volume (0-80 nm). (e, f) Height profiles (along the white arrow) of the quantitative reflection image acquired using (e) 500-550 nm bandpass filter and (f) 525-1000 nm bandpass filter.

119

Figure 3.9 Reflection images of an Aplysia neuron cultured on a flat gold surface acquired using (a) 500-550 nm bandpass filter and (b) 525-1000 nm bandpass filter. The scale bar on each image corresponds to 100 μm. The rectangular shape of cell body on the figure represents the presence of some glia stack on the left top corner of the neuron surface.

120

Figure 3.10 (a) AFM image of an Aplysia neuron cultured on a plasmonic crystal showing periodic structures on the surface due to the underlying plasmonic nanostructure. (b) 3D projection of the original AFM data of Figure 3.10a showing periodic topological features on the neuronal surface. (c) 3D projection of AFM data of Figure 3.10a after FFT filtering to remove plasmonic nanostructure contribution. (d) SEM image of a portion of an Aplysia neuron cultured on plasmonic crystal showing the nanostructured surface underneath the cell structures. The scale bar corresponds to 100 μm. (e) Index-corrected thickness (θ) calibration applied to the reflection image of an Aplysia neuron cultured on a plasmonic crystal acquired using 525-1000 nm bandpass filter. The scale bar corresponds to 100 μm. (f) AFM height profile along the white arrow drawn on Figure 3.10a. (g) Reflection contrast height profile along the white arrow drawn in Figure 3.10e.

121

Figure 3.11 Normalized reflection contrast calibration curves as a function of index-corrected thickness (θ) for fixed cell acquired using bandpass filters: 570-600 nm (yellow green triangles), 570-1000 nm (pink squares) and a combination of the 570-600 nm and 570-1000 nm data (dark cyan circles). A power law regression was applied to the combination calibration curve (red curve, NRC = 0.01565θ0.6323).

122

Figure 3.12 Spectral sensitivity of CCD camera over the 400-1000 nm wavelength range. The green region highlights the 570-600 nm wavelength range, while the green and pink regions combined represent the 570-1000 nm wavelength range. (Sony ICX085AL 2/3-inch Progressive Scan CCD Image Sensor with Square Pixel for B/W Cameras, Datasheet No. E95Z10C73; Sony Corp., http://www.alldatasheet.com/datasheet-pdf/pdf/47409/SONY/ICX085AL.html)

123

Figure 3.13 (a) Wavelength combined (570-600 nm and 570-1000 nm) reflection contrast calibration applied to wavelength combined image of an Aplysia neuron cultured on a plasmonic crystal. (b) Height profile along the white arrow drawn on the Figure 3.13a.

124

Figure 3.14 (a) Reflection image of Aplysia neurons cultured on a plasmonic crystal acquired using a 500-550 nm bandpass filter. The scale bar corresponds to 100 μm. (b) Rotated and magnified images of the region in the red box at Figure 3.14a acquired using 500-550 nm (left), 570-600 nm (middle), and 610-700 nm (right) bandpass filters. Based on interferometric estimates, the region indicated by the red asterisk and arrow in Figure 3.14b is ~330-400 nm thick.

125

Figure 3.15 Ray diagram for thin film interference from light reflected from two adjacent surfaces.

126

Figure 3.16 (a) FFT filtered AFM image of the Aplysia neuron structure displayed in Figures 3.14 (corresponding to the blue box in Figure 3.14b). (b) AFM height profile along blue arrow drawn on the Figure 3.16a.

127

Figure 3.17 (a) Thickness calibration applied to combined wavelength (570-600 nm and 570- 1000 nm) image of Aplysia pedal ganglion neurons on plasmonic crystal with a neurite (‘A’) and the remnants of the growth cone peripheral region (‘B’). (b) 500-550nm reflection image of an Aplysia pedal ganglion neuron on a plasmonic crystal with a neurite (‘C’) and the remnants of the growth cone peripheral region (‘D’ and ‘E’). (c) 500-550nm reflection image of an Aplysia pedal ganglion neuron on a plasmonic crystal. ‘Bulls-eye’ structures (denoted with red asterisks) correspond to the central region of the growth cone and are surrounded by thinner ‘veils’ corresponding to the peripheral region of the growth cone. The scale bar on (a) and (b) corresponds to 100 μm and that on (c) corresponds to 50 μm.

128

CHAPTER 4

QUANTITATIVE REFLECTION IMAGING FOR MORPHOLOGY AND DYNAMICS OF LIVE APLYSIA CALIFORNICA PEDAL GANGLION NEURONS ON NANOSTRUCTURED PLASMONIC CRYSTALS

4.1 Abstract

We demonstrate the application of a simple reflection imaging system which consists of a plasmonic crystal, a common laboratory microscope, and bandpass filters for quantitative imaging and in-situ monitoring of live cells and their interactions with the substrate. Surface plasmon resonance (SPR) is highly sensitive to changes in the local optical profile at the metal/dielectric interface. Changes in the thicknesses or dielectric properties of thin films near the surface of the plasmonic crystal are detectable by examining optical responses of plasmonic modes. Polyelectrolyte thin films deposited using the layer-by-layer (LBL) self-assembly method serve as a reference model system to verify reflection contrast changes when increasing the polyelectrolyte film thickness and changes associated with the optical responses originating from multiple plasmonic features. Finite-difference time-domain (FDTD) simulations allow the optical responses measured experimentally from the polyelectrolyte reference system to be adjusted for investigating cellular structure and functions. Living Aplysia californica pedal ganglion neurons immersed in artificial sea water (ASW) were used as a model system for this plasmonic reflection imaging technique. Here, the morphology of peripheral structures (structures < ~80 nm in thickness) of live Aplysia pedal neurons were quantitatively analyzed and cell detachment induced by trypsinization was visualized and interpreted. These results validate plasmonic reflection imaging as a technique that shows promise as a real-time analysis method for investigating the morphology and dynamics of ultrathin cellular systems in a complex environment.

129

4.2 Introduction

Cell morphology and dynamic biological processes are essential to understanding cell differentiation, growth and death in medical science and biotechnology.[1-6] Time-resolved imaging of biological events and structural changes inside of living cells provides insights into cell function that static observation of fixed cells would not reveal.[7-10] Because most cells are optically transparent, live cell imaging is difficult using conventional optics without staining or tagging the cell.[11, 12] Regardless of the imaging technique, cells’ health is crucial to preserve a normal metabolic state within live cell imaging experiments.[13] For imaging of cellular dynamics, the experimental conditions such as pH[14, 15], oxygenation[16], temperature[17] and osmolality[18] should be satisfied to maintain viability of live specimens. In addition, high intensity light illumination may induce temperature fluctuations and/or damage DNA via UV light absorption, ultimately causing cell death.[19] Therefore, it is critical that live cell imaging techniques are designed to minimize physical and chemical damage to the cell and suitable for the local biological environment.

Various optical analytical tools have been developed to study structure, composition, or vital phenomena of living cells. Fluorescence microscopy is the most widely used technique for live cell imaging because, by tagging cell components with fluorophores such as small organic dyes or quantum dots, it is capable of probing specific structures or materials inside of the cell.[20-22]

Fluorescence imaging, however, is complicated by the toxicity associated with introducing foreign fluorescent labels into the system and is also limited by photobleaching of the added fluorophores.[9, 23, 24] Phase contrast microscopy[25, 26] and differential interference contrast microscopy[27, 28] have been developed to visualize unstained, transparent biological specimens with enhanced contrast and resolution. Although these imaging methods are useful to study

130 cellular structures and processes, their two-dimensional (2D) data are insufficient for quantitatively analysis of the live cell morphology. Three-dimensional (3D) live cell images can be achieved using confocal microscopy – wherein the resultant images are reconstructed from multiple 2D optical slices obtained using different depths within a sample.[29-32] Confocal microscopy also requires incorporation of fluorescent labels in the cells, which introduces the aforementioned problems and increases the overall complexity of the experiment.[33] Digital holographic microscopy[12] and spatial light interference microscopy[34] are able to provide a label-free form of quantitative live cell imaging yielding information about morphology and dynamics of the cell, but these optical techniques need complex and demanding setups as well as non-trivial data processing. Atomic force microscopy (AFM) is an alternative to optical imaging techniques that can be used to generate a topographic map of specimens with high resolution in the perpendicular direction.[35-38] While AFM offers delicate height data at nanometer resolution, it is not suitable for studying a live cell dynamics due to the long scan times and the unavoidable physical contact between the AFM tip and the sample.[39]

To overcome the drawbacks of conventional cell imaging methods, we previously developed a label-free reflection imaging technique using a plasmonic crystal which quantitatively maps the thickness of complex biological structures in dried condition (fixed Aplysia californica pedal neurons).[40] This plasmonic imaging method is label-free and highly sensitive to changes on the surface of the plasmonic crystal because surface plasmon resonances (SPRs) generate an enhanced evanescent electric field at the interface of a metal/dielectric. Nanoimprinted plasmonic crystals used as a substrate in this system allow for a simple and inexpensive method for imaging live cells events when combined with a simple imaging platform consisting of a common laboratory microscope and a charge-coupled device (CCD) camera.

131

Herein we extend the application of this analytical technique to visualize the morphology and dynamics of live cells. The polyelectrolyte layer-by-layer (LBL) assemblies were used as reference model to calibrate contrast changes as a function of thickness of the thin film on the surface of a plasmonic crystal. Theoretical considerations examined using finite-difference time- domain (FDTD) simulations help to establish a contrast calibration for the quantitative analysis of complex biological system from reference contrast data. Wavelength-dependence of contrast changes is confirmed experimentally using bandpass filters to roughly restrict wavelength range collected for imaging. This multispectral reflection contrast calibration was applied to the reflection images of live Aplysia pedal neurons on the surface of a plasmonic crystal to demonstrate that this plasmonic reflection imaging technique can quantitatively map the thickness of peripheral regions of live cells in a complex biological environment. We further show this imaging method can be utilized to visualize and analyze the interactions occurring between a plasmonic crystal and a biological system by examining contrast changes of reflection images of live Aplysia pedal neurons associated with trypsin-mediated cell detachment.

4.3 Experimental Materials and Methods

4.3.1 Materials

Reagents were used as received without further purification. Photocurable polyurethane

(NOA73) was purchased from Norland Products. Polydimethylsiloxane (soft PDMS; Sylgard

132

(SIP6831.1) and (1,3,5,7-tetravinyl-1,3,5,7-tetramethylcyclotetrasiloxane) (7900) were purchased from Gelest. Manganese chloride tetrahydrate, poly(sodium 4-styrenesulfonate) (PSS,

MW = 70,000 g/mol), poly(allylamine hydrochloride) (PAH, MW = 70,000 g/mol), 16- mercaptohexadecanoic acid (MHDA), and trypsin (T1426) were purchased from Sigma-Aldrich.

Ultrapure water (18 MΩ) was generated using a Millipore Milli-Q Academic A-10 system and used to prepare the polyelectrolyte solutions and artificial sea water (ASW).

4.3.2 Plasmonic Crystal Fabrication via Soft Nanoimprint Lithography

Full-3D plasmonic crystals were fabricated using soft nanoimprint lithography technique as previously reported.[40-45] First, a double-layer PDMS stamp (composite hard-PDMS/soft-

PDMS) was cast from a patterned photoresist master consisting of nanohole array relief structures and subsequently used to fabricate plasmonic crystal. Several droplets of the NOA prepolymer was cast onto a glass slide by pressing PDMS stamp into the liquid pre-polymer and then cured by exposing to ultraviolet light for ~20 min. After carefully peeling off the PDMS stamp, the NOA nanostructures were usually placed in a 70 °C overnight to complete curing with heat treatment. The replicated nanostructures were a uniform square array of nanoholes with a hole spacing (center to center) of ~740 nm, a hole diameter of ~540 nm, and a relief depth of

~285 nm. After a sputtering to deposit ~5 nm of titanium dioxide adhesion layer on top of the

NOA nanohole array, the ~45 nm gold was deposited by sputter deposition in 5 mTorr argon

(AJA International). To improve cell outgrowth on a plasmonic crystal and prevent water penetration into the interface between metal film and NOA nanostructures, a thin aluminum oxide of ~6 nm was deposited with atomic layer deposition (Cambridge Nanotech).

133

4.3.3 Transmission Measurements of Plasmonic Crystals.

Transmission spectra of plasmonic crystals were measured using a Varian 5G UV-Vis-NIR spectrophotometer with normal incident light and no temperature control. Transmission spectra over a wavelength range of 355-1200 nm were collected and used to verify the geometry of plasmonic crystals for further computational modeling.

4.3.4 Growth of Polyelectrolyte Layer-by-Layer Assemblies on Gold Films for Reflection

Imaging Contrast Calibration

Polyelectrolyte assemblies were created on the surface of plasmonic crystal or gold-coated pieces of silicon wafer by a LBL method. To facilitate a LBL deposition, carboxyl-terminated self- assembled monolayers were formed on the gold film surface by immersing the substrates in an ethanol solution of MHDA (2 mM) for at least 18 h, after which the substrates were rinsed thoroughly with ethanol and dried under nitrogen flow. Layers of PAH and PSS were then alternately deposited onto the surface of gold coated substrate. The substrate was immersed in a

PAH solution (3 mg PAH/mL in water, pH = 8.0) for 5 min, rinsed thoroughly with water and dried with nitrogen gas. This PAH-layer terminated substrate was then immersed in a PSS solution (3 mg PSS/mL in 1M MnCl2, pH = 2.0) for 90 s, rinsed with water and dried under nitrogen flow.

4.3.5 Thickness and Refractive Index Measurements of Polyelectrolyte Layer-by-Layer

(LBL) Assemblies in Artificial Sea Water (ASW) with Atomic Force Microscope (AFM) and Ellipsometry

The thicknesses of polyelectrolyte LBL assemblies on gold-coated silicon in ASW were measured using an Asylum Research MFP-3D atomic force microscope and the refractive index

134 of polyelectrolyte films under the same conditions was determined using a Woollam VASE spectroscopic ellipsometer with home-built liquid cell (shown in Figure 4.1). For simple analysis, gold-coated silicon wafer pieces were used for AFM and ellipsometry measurement after surface modification with a MHDA monolayer. AFM can directly measure the thickness of polyelectrolyte layer in ASW. Areas of the gold-coated silicon were masked using permanent marker to dictate the pattern of polyelectrolyte LBL assemblies. The permanent maker mask can be easily removed by rinsing with acetone, and then the polyelectrolyte thickness was determined by measuring the edge step at the interface of the polyelectrolyte covered and non- covered regions via AFM. AFM measurement was operated in tapping mode under the liquid environment (in ASW) using soft cantilever with 0.08 N/m spring constant which was purchased from Asylum Research. Analysis of the AFM data was performed using IgorPro. The thickness data of polyelectrolyte film acquired from AFM measurement was employed to determine refractive index of polyelectrolyte film via ellipsometer. The home-built liquid cell was designed and manufactured to evaluate the optical properties of polyelectrolyte film in ASW using ellipsometer. At every second round of polyelectrolyte deposition, ellipsometry data were collected over the wavelength range of 300-900 nm at an incident angle of 60°. The polyelectrolyte layer was modeled as a Cauchy material. The parameters of Cauchy model for polyelectrolyte layer were theoretically determined based on ellipsometry spectrum and thickness value of polyelectrolyte film, which is used to calculate the refractive index of polyelectrolyte assemblies.

4.3.6 Reflection Imaging of Plasmonic Crystals Using a Laboratory Optical Microscope

Reflection mode images of both the polyelectrolyte films and cells cultured on the plasmonic crystal surface were obtained using a Zeiss AxioScope A1 with halogen light source and a 20×

135 and 0.40 NA of objective lens. Bandpass filters are inserted in the microscope immediately in front of the microscope camera (AxioCam MRc). Images were captured using an AxioVision with 1388 × 1040 array of pixels (each pixel 6.5 × 6.5 μm). The exposure time was varied to control the overall exposure, while the halogen lamp intensity was fixed in low level to reduce possible thermal damages to live cells.

Grayscale reflection images were processed and analyzed using Matlab and ImageJ. Vignetting in the images of polyelectrolyte layer and live cell was corrected using a pixel-by-pixel division of the experimental image by the reference mirror image which is the image of original plasmonic crystal taken with the corresponding bandpass filter. With assumption that the contrast values of each pixel have standard normal distribution, pixel values out of 80% confidential intervals were considered as blemishes or debris and removed from the reflection imaging contrast calibration.

4.3.7 Reflection Contrast Calibration

Part of MHDA-coated nanohole array was covered with cured NOA film to create a region at which the surface refractive index was maintained constantly during the polyelectrolyte film deposition. The reflection images of this plasmonic crystal were taken at every second round of polyelectrolyte depositions to calibrate the normalized reflection contrast changes as a function of thickness of polyelectrolyte assemblies. The normalized reflection contrasts of polyelectrolyte coated regions (PEL regions) were determined through the two-step process, as we reported previously.[40] In the first, the average pixel intensities of the optically constant regions (NOA regions) in all images were equated to correct the differences in illumination intensity between images using Equation (4.1):

136

Reference NOA average Scaled PEL average =  Image PEL average (4.1) Image NOA average where the reference average is defined as the average pixel intensity of reflection image for the plasmonic crystal before the polyelectrolyte deposition. Then, the scaled average PEL pixel values were normalized by the reference PEL average according to Equation (4.2):

[scaled PEL average] - [reference PEL average] Normalized reflection contrast = (4.2) [reference PEL average]

4.3.8 Cell Culture of Aplysia californica Pedal Ganglion Neurons

Sea slug Aplysia Californica (100-300g) were supplied by the National Resource for Aplysia

(Miami, FL) and kept in circulated, aerated seawater at 14 °C. Prior to dissection, the animals were anesthetized by injection of isotonic magnesium chloride solution into the body cavity

(~30-50 % of body weight). Individual Aplysia pedal neurons were isolated after incubation in

ASW (in mM; 460 NaCl, 10 KCl, 10 CaCl2, 22 MgCl2, 26 MgSO4, and 10 HEPES, pH 7.8) with proteases (1% type XIV, Sigma-Aldrich) at 34 °C for between 60-120 min as described previously.[46, 47] A poly-L-lysine layer (1-2 nm) was formed on the plasmonic crystal surface.

The cells were mechanically isolated in ASW, transferred onto the plasmonic crystal surface immersed in ASW supplemented with antibiotics (ASW containing 100 units/ml penicillin G,

100 mg/ml streptomycine, and 100 mg/ml gentamicin, pH 7.7) and left to attach and grow overnight at room temperature. Six to eight pedal neuron cells were cultured on the plasmonic crystal surface at a time.

137

4.3.9 Atomic Force Microscopy of Fixed Neuron Cells on Plasmonic Crystals

Due to the limitations on live cell imaging using AFM, cultured Aplysia neurons on plasmonic crystals were dried and fixed before AFM measurements. Cells were fixed by addition of 1 mL of 4% paraformaldehyde to 3 mL ASW antibiotic culture media and occasional stirring for 30 s, removal of 1mL of the solution, addition of another 1mL of the 4 % paraformaldehyde solution and exposure for 30 s, followed by the removal of all solution. After removal of the fixation solution, the cells were carefully rinsed with deionized water and dried. Height profiles of fixed

Aplysia neurons on the surface of the plasmonic crystal were measured using an Asylum

Research MFP-3D atomic force microscope operated in tapping mode. Analysis of the AFM data was performed using IgorPro.

4.3.10 Cell Dissociation from Plasmonic Crystal Surface Using Trypsinization

Cell detachment was induced by injection of trypsin into ASW culture media. Trypsin was diluted in ASW to reach 10 mg/mL of concentration. For the detachment of Aplysia neuron cell from the substrate, 0.3 mL of trypsin solution was injected into 3mL of ASW antibiotic culture media in which live cell cultured plasmonic crystal was immersed.

4.3.11 Finite-Difference Time-Domain (FDTD) Simulations of Plasmonic Nanostructures

3D FDTD simulations were carried out to model the 0th-order reflection/transmission spectra

(with light normally incident on the glass substrate) and to verify optical behavior changes as a function of thin film thickness on a plasmonic crystal using commercial FDTD software package

(Lumerical FDTD, Lumerical Solutions Inc.). The unit cell geometry defined a gold nanohole in the x-y plane with a 740 nm center-to-center hole spacing, 544 nm hole diameter, 280 nm relief depth, 44 nm Au film on the top surface of the plasmonic crystal, 16 nm Au film conformally

138 coating the nanohole sidewalls, and 16 nm Au film on the bottom of the nanoholes. The frequency dependent dielectric constant of gold was modeled using previously reported parameters from a Drude plus two-pole Lorentzian model.[36] The refractive indices of NOA, polyelectrolyte, refractive index-corrected material for live cell, and ASW were taken to be 1.56,

1.4 8, 1.42, and 1.34, respectively.

4.4 Results and Discussion

4.4.1 Plasmonic Crystals and Computational Models

We previously developed a simple and powerful analytical imaging technique using plasmonic crystal to quantitatively monitor the morphology of a biological system, the fixed pedal ganglion

Aplysia californica central neuron.[40] This plasmonic reflection imaging method has great potential as an analytical tool in biology fields because it can be operated under the complex environment (i.e. in the liquid cell media) without complicated optical instrumentation. Herein we prove the capability of this plasmonic reflection imaging technique by quantitatively studying the morphology of live Aplysia californica neurons and cell detachment stimulated by trypsinization. The optical setup of this imaging technique is illustrated in Figure 4.2. Reflection images of live Aplysia neurons on a plasmonic crystal immersed in liquid cell media (ASW) were acquired with a common laboratory microscope and bandpass filters inserted in front of a

CCD camera.

Plasmonic crystals similar to those we used for previous study with fixed cells[40] was employed as a substrate for live cell imaging. The plasmonic crystals were fabricated using a soft nanoimprint lithography method that provides robust and cost-effective route for generating

139 highly uniform nanostructures over large area. The inset of Figure 4.2 shows an illustration of cross section of full-3D plasmonic crystal consisting of a square array of embossed nanoholes

(545 nm diameter, 280 nm depth, and 740 nm pitch) conformally coated with a thin film of Au.

This plasmonic crystal can generate multiple plasmonic features that are highly sensitive to optical profile changes on metal/dielectric interface at visible wavelengths.[41-45] The complex optical features of plasmonic crystals are supported by FDTD calculations with appropriate boundary conditions and geometrical model of periodic nanohole array. Figure 4.3a presents a structural model of a full-3D plasmonic crystal. The geometric parameters used for modeling the plasmonic crystal are verified by the strong agreement between the experimentally measured and the calculated transmission spectra (displayed in Figure 4.3b).

4.4.2 Index Corrected Mass Coverage

The polyelectrolyte LBL assemblies with PAH and PSS were used as a well-defined reference system to calibrate the reflection contrast changes as a function of thickness of thin film on plasmonic crystals. For further theoretical analysis using FDTD calculations, the accurate thicknesses and refractive index of the polyelectrolyte layer in ASW are required. The thickness of one layer of PAH/PSS and refractive index of polyelectrolyte system in dry condition were determined by previous studies,[40, 44, 48] but these values must be altered in the ASW environment. Therefore, we measured increasing thickness of polyelectrolyte assemblies using

AFM, and then determined refractive index polyelectrolyte film using ellipsometry.

Spectroscopic ellipsometry is useful optical technique for exploring the dielectric properties of thin films, but it is difficult to determine the complex refractive index without a rough estimate of specimen thickness, as the ellipsometric signal depends on both of the thickness and optical

140 properties of thin film. Therefore, AFM was used to measure the thickness of polyelectrolyte film patterned on a gold-coated silicon wafer immersed in ASW. To pattern the polyelectrolyte film on flat silicon wafer, a patterning mask was drawn using a permanent marker and removed by rinsing with acetone after polyelectrolyte deposition (illustrated in Figure 4.4). Figure 4.5a shows a comparison of ellipsometric responses of polyelectrolyte film measured before and after acetone rinsing, demonstrating that the polyelectrolyte film is not negatively affected by acetone.

The steep step edge of the polyelectrolyte pattern (shown in Figure 4.5b) can be easily quantified using AFM measurement. These AFM data reveal that each PAH/PSS deposition step increases the thickness of the polyelectrolyte assembly by 3.85 ± 0.043 nm in ASW.

With the thickness data of polyelectrolyte assemblies and Cauchy material model defining optical properties of polyelectrolyte film, the refractive index can be determined using spectroscopic ellipsometry. The ellipsometric parameters of the polyelectrolyte assemblies were determined to be An = 1.477, Bn = 0.001, and Cn = 0.0005 such that the refractive index at 600 nm was calculated to be 1.484 using the Cauchy equation, in agreement with previous reports.[49] The thickness increase due to the each PAH/PSS deposition step determined by

AFM measurement and ellipsometric analysis is plotted in Figure 4.6. As the volume change of polyelectrolyte assemblies in ASW originates from ASW penetration into thin film, the refractive index of polyelectrolyte in ASW can be approximated using the Arago-Biot equation,

n1 n 1 2 n 2 where n is the refractive index of polyelectrolyte assemblies in ASW, n1 is the

refractive index of dried polyelectrolyte assemblies (1.64), n2 is the refractive index of ASW

(1.34), 1 is the volume fraction of the polyelectrolyte assemblies, and 2 is the volume fraction of ASW. According to this relationship, the calculated refractive index of the polyelectrolyte

141 film is 1.485, which is in good agreement with experimentally measured refractive index of polyelectrolyte assemblies in ASW using spectroscopic ellipsometry.

SPR response is dependent on both the local refractive index and the changes in thickness of dielectric material on the plasmonic crystal. A living cell is a complex structure, which contains many organelles with different refractive index, including membrane, organelles, cytoplasm, nucleus, etc. For example, the refractive index of the cell membrane (1.46-1.53) is normally higher than that of the cytoplasm (1.35-1.38). Since the optical properties differ between the polyelectrolyte assemblies and biological systems, the interpretation of the imaging mode contrast must account for the local refractive index of the cellular structures. This quantitative reflection imaging technique is limited to a thickness range of approximately 80 nm.[40] At least quarter of the cellular component is expected to be the cell membrane (for reference, thickness of cell membrane is ~ 10 nm [50, 51]). Therefore, using the Arago-Biot approximation, we estimated the refractive index of biological material in our system is 1.42.

Finite-difference time-domain (FDTD) calculations were used to determine the index corrected mass coverage (θ) for a biological material (n ~ 1.42) that exhibits the same optical response as the polyelectrolyte assemblies (n ~ 1.48) at different thicknesses. Figure 4.7 presents the calculated reflection spectra for the plasmonic crystals immersed in ASW and coated with a conformal layer of either polyelectrolyte or the equivalent index corrected material for live cells.

The small discrepancies between the spectra likely originated from the limitations in the grid spacing of 4 nm for structure modeling. Those FDTD calculated corresponding pairs for polyelectrolyte film thickness and index corrected mass coverage are presented in Figure 4.8.

142

The cell membrane is usually separated from the substrate and cell adhesion to substrate is supported by extracellular matrix, a collection of fibrous protein and glycosaminoglycan.

According to the previous study, the distance between cell and substrate at peripheral lamellipodia was determined to be 25  10 nm.[52] To demonstrate the effects of gaps between cell and substrate on reflection spectra, three models were considered; plasmonic crystals with uniform index corrected mass coverage and those with the same index corrected mass coverage with either a 12 nm or 24 nm ASW gap between the plasmonic crystal substrate and the index corrected material. The calculated reflection spectra of those three models are present in Figure

4.9 and show a negligible effect of a thin ASW gap (~ 24 nm) on plasmonic crystal reflectance in visible light range. The effect of gap between cell and substrate on reflection images may become smaller than that we estimated here using three different models since the refractive index of extracellular matrix is larger than that of ASW. Thus, we did not consider the gap between cell and substrate for further quantitative analysis for thickness of live cells.

4.4.3 Reflection Imaging Contrast Calibration

Reflective images of polyelectrolyte films deposited on plasmonic crystals immersed in ASW were acquired using a common laboratory optical microscope with bandpass filters inserted directly in front of the camera to control the collected wavelength for imaging. The optical setup for live cell imaging is illustrated in Figure 4.2. The NOA covered region where the refractive index are constant was used to correct the variation in illumination intensity between reflection images, and images of plasmonic crystals obtained before the polyelectrolyte deposition were defined as reference images to normalize the contrast change as a function of polyelectrolyte film, in the same manner as we previously reported.[40] One difference between this and the previous

143 method is that here the plasmonic crystal was imaged from the front side. A representative image of a plasmonic crystal used for contrast calibration is presented in Figure 4.10a.

Figure 4.10b shows the step-edge profile of the normalized reflection contrast at the boundary

(along the yellow arrow shown in Figure 4.10a) and the contrast difference between the polyelectrolyte-coated and NOA-coated regions. This step-edge profile was fitted using a cumulative Gaussian curve[53] (red line, as presented in Figure 4.10b) to estimate the lateral resolution of the plasmonic imaging sensor. This resultant fit was well modeled with a Gaussian width of 0.84 µm corresponding to the spatial resolution, a value that is slightly larger than the resolution limit of CCD camera and microscope used for this study, ~0.65 µm . As discussed in previous report,[40] a single pixel of reflection image is overlapping part of one or several nanoholes because the pixel size (~0.65 µm ) and periodicity of nanoholes (~0.74 µm ) are comparable. Therefore, the pixel intensity of each pixel in both regions on a plasmonic crystal slightly varies depending on nanohole coverage, even though the nanoholes are not resolved in the reflection images.

By combining the results of corresponding pairs of polyelectrolyte and index corrected material

(described in previous section) and contrast changes as the thickness of the polyelectrolyte assemblies increases, the relation of normalized reflection contrast (NRC) and θ was established

(displayed in Figure 4.11). The error bars for contrast data are obscured because large numbers of pixels (~ 30,000) are used to determine the average contrast. The reflection contrast increases with θ in the wavelength ranges of 570-600 nm and 570-1000 nm, while the reflection contrast decreases and becomes negative at the wavelength ranges of 500-550 nm and 600-700 nm. From these varying contrast changes obtained using different bandpass filters, we can conclude that contrast inversion on reflection images under the different wavelength ranges is attributed to

144 different effective plasmonic features depending on wavelength ranges of light. The normalized reflection contrast changes over the wavelengths of 600-700 nm and 570-1000 nm are non- linearly changed, while those measured using 500-550 nm and 570-600 nm shows simple linear regression. These non-linear complex behaviors of reflection contrast are considered to be the results of multiple plasmonic modes generated by squared array of nanoholes. For simple interpretation, linear normalized reflection contrast changes for both wavelength ranges of 500-

550 nm and 570-600 nm were used in following sections: NRC  0.00089  0.00218for the

500-550 nm bandpass filter, and NRC 0.00138 0.00175for the 570-600 nm bandpass filter.

The reflection spectra of a plasmonic crystal (Figure 4.12a) and a flat gold substrate (Figure

4.12b) coated with increasing θ in ASW were calculated using FDTD computational method to theoretically confirm optical response changes over the range of wavelengths. The FDTD calculated reflectance on the flat gold surface decreased monotonically over the wavelengths of

400-1000 nm as θ increases, whereas the reflection spectra of plasmonic crystal under the same conditions show more complex and wavelength-dependent optical behaviors. The variations in the plasmonic crystal reflectance stems from the multiple plasmonic features and combinations of those on plasmonic crystal. For example, wavelength ranges of 600-740 nm exhibit a decrease in reflectance with thicker θ while the opposite trend is observed in the 770-800 nm range. These trends support the experimentally observed wavelength-dependent contrast inversion of reflection images shown in Figure 4.11. Using FDTD calculations of reflection spectra, we can also estimate the sensing volume of our plasmonic sensor platform. Figure 4.12a shows that the reflection spectrum of plasmonic crystal with 80 nm of θ is almost identical with that of plasmonic crystal with 100 nm of θ. Therefore, the quantitatively measurable range of thin film on plasmonic crystal in our SPR-based system is less than 80 nm from the surface.

145

Although calculated reflection spectra of plasmonic crystal provide insight into the complicate optical behaviors under different wavelength ranges, these computational results are not enough to describe the contrast changes on reflection images. The simulation assumes that the incident light is normal to the surface. Experimentally, the illumination angle of light radiated through an objective lens with a numerical aperture of 0.40 should be more than zero (maximum angle of

24°), which significantly influences the optical behavior of plasmonic crystals. We also did not consider the light refraction at the interface of air and ASW by using the modeled light source located in ASW. This assumption reduces the complexity and computation time, but causes some deviation from the experimentally obtained results. More intensive modeling may be required to fully analyze the optical behavior of the plasmonic crystal under the current optical setup.

4.4.4 Reflection Imaging of Live Aplysia Pedal Ganglion Neurons on Plasmonic Crystals

To explore the capability of SPR-based reflection imaging technique using plasmonic crystals as a quantitatively analytical method for complex biological material in the non-specific condition, live Aplysia pedal ganglion neurons were cultured on the surface of a plasmonic crystal immersed in ASW and imaged using our optical setup for further discussions. Figure 4.13a and

Figure 4.13b show the reflection images for part of a live Aplysia neuron acquired using different bandpass filters (original images are shown in Figure 4.14); Figure 4.13a collected using a 500-

550 nm bandpass filter, and Figure 4.13b collected using a 570-600 nm bandpass filter. The peripheral regions of the live neurons appear darker than surrounding area in the 500-550 nm image (Figure 4.13a) while those in the 570-600 nm image appear brighter than surrounding area

(Figure 4.13b). The contrast inversion estimated by results of normalized reflection contrast changes acquired using different bandpass filters (Figure 4.11) can be verified from those two images.

146

Although Aplysia neurons are consisted with various features with range in thickness from tens of nanometers to several microns, only peripheral area can be quantitatively imaged using our

SPR-based technique because this system is only sensitive to range of 80 nm from the surface of plasmonic crystal. Therefore, we focus only on very thin features whose dimensions are within

80 nm using the linear regression of optical response as a function of θ observed at 500-550 nm and 570-600 nm wavelength ranges.

The linear contrast calibration data presented in Figure 4.11 was applied to each reflection images shown in Figure 4.13a and Figure 4.13b to analyze thickness of live Aplysia pedal neurons. The normalized reflection contrast was calculated through mathematically dividing each pixel in the cell by the average pixel value of reference area, as described above in the

Experimental section. The reference area was selected to be in a location where cell features were absent but adjacent to the cell. The “empty” reference regions in the image lacked cell features, but exhibited a thin film of poly-L-lysine, which was coated prior to cell transfer to improve the biocompatibility of plasmonic crystal and help neuron cells to attach and grow on surface of substrate. Because the poly-L-lysine layer was applied to the entire plasmonic crystal area, the contributions of this layer on reflection contrast can be neglected when normalizing the reflection contrast. Figure 4.13c and Figure 4.13d show the resultant thickness-transformed cell images using the 500-500 nm bandpass filter and 570-600 nm bandpass filter, respectively.

In the thickness-transformed cell images, the thickness of live Aplysia pedal neurons were restricted between 0 and 80 nm due to the limitation of sensing volume of SPR-based imaging technique. Pixels with values outside of the range of reflection contrast corresponding to cell thicknesses from 0 to 80 nm were truncated and assigned to minimum and maximum contrast values. The cell body of Aplysia pedal neurons is impossible to analyze with this technique as the

147 features are far thicker than the plasmonic sensing volume. The contrast of thicker features than the 80 nm of sensing volume is the result of convolution of various optical phenomena such as light absorption, reflection, and scattering by cellular structure, and interference effects as well as

SPR. These multiple optical effects cause the thicker neuron body to appear dark on both images of Figure 4.13a and Figure 4.13b obtained using different bandpass filters, while thinner outgrowths and growth cones show contrast inversion depending on wavelength ranges used to collect each image. From the thickness-transformed images from both wavelength regions

(Figure 4.13c and Figure 4.13d), the thickness of thin features of live Aplysia pedal neurons are roughly identified; the thickness of thin regions near the cell body are ~10-20 nm and that of string-like filaments are ~60 nm or more.

The self-consistency of estimated thicknesses calculated using different bandpass filters (Figure

4.13c and Figure 4.13d) proves the reliability of this thickness evaluating protocol. Figure 4.13e and Figure 4.13f show almost identical thickness profiles along the white arrows on the cell images of Figure 4.13c (using 500-550 nm bandpass filter) and Figure 4.13d (using 570-600 nm bandpass filter). The thickness values of the filament features located along the white arrows are in good agreement; thickness of ~70 nm, ~60 nm, and ~30 nm at X ~8 μm, ~16 μm, and ~20 μm, respectively. The correspondence on thickness data becomes poorer in the very small features; for example, the small feature at X ~ 13 μm is observed in 500-550 nm image (Figure 4.13c), but not in the 570-600 nm image (Figure 4.13d). The discrepancies of those two images are estimated to be originated from the relatively coarse linear regressions and reference contrast calculation. An underestimation or overestimation of calculated reflection contrast of thin features can arise from these two influences. As demonstrated from previous work[40], combination of multispectral reflection images can increase the sensitivity for small features and

148 contrast of image. In addition, more complex fitting for reflection contrast changes as a function of thickness may help to improve this imaging technique for quantitative analysis.

Atomic force microscopy (AFM) was also adopted to evaluate the reliability of this SPR-based analysis method for quantifying thickness of small features in the neuron cells. AFM is a powerful tool to directly map the topography of a surface, while its inevitable physical contact between AFM tip and the surface makes it inappropriate for measuring live cells due to viability issues. In our trials, it was very difficult to measure the heights of specific parts of live Aplysia pedal neuron without damaging the cell. Due to this issue, neuron cells were fixed onto the plasmonic crystals for AFM scanning after imaging experiments. The height of features in fixed cell should be reduced compared to that in live cell due to the dehydration. We estimated the height of live neuron cell from AFM height data of fixed neuron based on the fact that about 70% of the cell volume is composed of water. If the height shrinkage was assumed as a result of water removal only, the effect of fixation should be a 33% decrease in thickness of cell height.

Although there is a loss in resolution, the periodic modulation in the AFM images stem from the underlying plasmonic crystal was filtered out using a Fast Fourier Transform (FFT) of the image to clearly show height of cell without effects of the substrate, in agreement with our previously reported results.[40]

The FFT filtered AFM image of the fixed neuron cell obtained by scanning the region around the white arrow in reflection images of identical live Aplysia pedal neuron (shown in Figure 4.13) is presented in Figure 4.15a. The height profile of the fixed cell along the white arrow in AFM image (Figure 4.15a) was converted to the estimated thickness of live cell using shrinkage assumption. Those two height profile data are shown in Figure 4.15b. The height profile of the estimated live cell (presented in Figure 4.15b) reveals good agreement with height profiles

149 calculated from the reflection images using different bandpass filters (presented in Figure 4.13e and Figure 4.13f). Height data of three string-like filaments are almost identical in the reflection and AFM images. Although there are good agreements in height of cell features, the spatial sizes of the three filaments in the AFM image are larger than in the reflection images. This different may result from relatively low spatial resolution of reflection image (~1 μm) compared to that of the AFM image. Despite this slight discrepancy, the AFM results offer independent verification of quantitative results obtained from plasmonic reflection imaging.

4.4.5 Imaging the Effect of Trypsin on the Cell Activity

The SPR imaging can be used to investigate the interactions between cell and substrate, as the evanescent field excited by surface plasmons is highly sensitive to changes close to the interface of metal and dielectric (~100-200 nm). Consequently, our plasmonic reflection imaging technique is capable of monitoring the influence of chemical stimulations on cell dynamics occurring near the surface of plasmonic crystal. Trypsin is a pancreatic serine protease with substrate specificity based upon positively charged lysine and arginine side chains. Because of its ability for orderly and unambiguous cleavage of protein molecules, trypsin is widely used to detach adherent cells from the substrate surfaces.[54-56] Cell detachment, one of the most important vital activities to understand cell growth and movement, can be experimentally monitored in real time and theoretically understood by comparing reflection images of live

Aplysia pedal neurons obtained before and after trypsinization.

Figure 4.16 shows the changes on reflection contrast of live Aplysia pedal neurons due to the injection of trypsin into culturing media. The 570-600 nm bandpass filter was used for imaging because its steep and linear response in contrast as a function of θ allows for higher quantitative

150 sensitivity. The contrast changes were not easily detected in the original images, but the mass coverage transformed image obtained by converting normalized contrast to normalized index mass coverage provides a quantitative method of monitoring the contrast change. The distinguishable changes on θ in mass coverage transformed images (presented in Figure 4.16) are not observed during the ~30 min from the trypsin treatment. This time delay of contrast changes on reflection images represents the onset time of trypsin activation. After 30 min from the trypsin stimulation, reflection contrast gradually changed. For example, an outgrowth filament readily distinguishable in t  0 min image (labeled ‘i’ presented in Figure 4.16a) is not observed in images acquired 45 min later from the applying trypsin and color of other outgrowth filaments

(in square of ‘ii’ and ‘iii’ presented in Figure 4.16a) changes from white-yellow to yellow-red, which means that the reflection contrast of cell in 570-600 nm images decreases due to the cell detachment.

After the trypsin is activated, the cells are detached from the substrate and the ASW layer is created between the substrate and the cell. The reflection spectra are blue shifted when refractive index near the surface of plasmonic crystal decreases. The reflection contrasts are expected to be reduced, which result from the inserted ASW layer having smaller refractive index relative to the cellular materials. Comparing two mass coverage transformed images of live Aplysia pedal neuron taken before trypsin was applied ( min, presented in Figure 4.17a) and after trypsin was activated ( t  60 min, presented in Figure 4.17b) demonstrates this anticipation of a decrease in reflection contrast. Figure 4.17c shows changes in mass coverage profiles along the white arrows on cell images due to the trypsinization (Figure 4.17a and Figure 4.17b). The smaller θ profile values obtained from min reflection image intuitively validate the

151 generation of ASW layer between cell and plasmonic crystal. Therefore, cell detachment can be simply monitored from the 2D reflection image using SPR-based optical system in real time.

Additional FDTD calculations of reflection spectra were performed to quantitatively investigate the ASW layer between cell and plasmonic crystal formed during cell detachment. In Figure

4.17c, the θ of outgrowth filament (indicated with red asterisk) is reduced from ~60 to ~35 as a result of trypsinization. Although average cell/substrate distance at peripheral region of the growth cone is 10-30 nm, the membrane of other parts of cell is usually separated from the substrate by 100-150 nm.[52] By cleaving contact between the cell and the plasmonic crystal using trypsin, ~150 nm of ASW layer may be generated between cells and substrate, which is supported by calculated reflection spectra using FDTD simulation (present in Figure 4.17d). The reflectance of plasmonic crystal having 160 nm of ASW gap between plasmonic crystal and 60 nm of index corrected material (160 nm ASW gap + 60 θ) behaves similar to that of a plasmonic crystal coated with 36 nm of index corrected material (36 θ) in the wavelength range used for imaging (570-600 nm). This good agreement between computed reflection spectra and θ changes on reflection images demonstrates that cell detachment can be quantified using reflection images and FDTD calculation. Therefore, our SPR-based quantitative reflection imaging is promising as an analytical technique to explore the cell dynamics occurred at interface between cell and substrate.

4.5 Conclusions

We demonstrate a plasmonic imaging technique using living pedal ganglion neurons of

Aplysia californica that is capable of quantitatively scrutinizing the morphology of nanoscaled

152 thin film structure and the dynamics occurring at the surface of a plasmonic crystal under a complex biocompatible environment. Thickness-controlled polyelectrolyte film deposited by layer-by-layer assembly method was used as reference system to evaluate the reflection contrast changes as a function of thickness of thin films on the plasmonic crystal. FDTD computations facilitated the comparison of polyelectrolyte film thickness to the index corrected mass coverage for the biological specimen, establishing contrast calibration for quantitative analysis on thickness of thin outgrowth region of live cell and interaction between cell and substrate.

Wavelength-dependency of contrast changes on reflection image is experimentally validated using bandpass filters which limit the range of collected wavelength for imaging. The thickness of live Aplysia pedal neurons was determined by applying contrast calibration to reflection images was confirmed using atomic force microscopy. Trypsin-mediated cell detachment from the plasmonic crystal was visualized and quantified using label-free plasmonic reflection imaging without negatively affecting cell viability. This biocompatible reflection imaging technique represents a powerful analytical method for studying the morphologies of ultrathin biological structures and dynamic interactions occurring on the surface of plasmonic crystals in real time. This plasmonic imaging technique has the potential to be utilized for quantitative monitoring and investigation of changes on morphology and chemical/physical activities of live cells while maintaining their vital functions.

4.6 Acknowledgements

153

Frontier Research Center on Light-Material Interactions in Energy Conversion and was performed using resources at the Frederick Seitz Materials Research Laboratory Central

Facilities at the University of Illinois, including the Center for Microanalysis of Materials; the

Micro/Nanofabrication Facility; and the Laser and Spectroscopy Facility.

154

4.7 References

1. Clarke, P.G.H., Developmental Cell-Death - Morphological Diversity and Multiple Mechanisms. Anatomy and Embryology, 1990. 181(3): p. 195-213. 2. Hogeweg, P., Evolving mechanisms of morphogenesis: on the interplay between differential adhesion and cell differentiation. Journal of Theoretical Biology, 2000. 203(4): p. 317-333. 3. Vogel, V. and M. Sheetz, Local force and geometry sensing regulate cell functions. Nature Reviews Molecular Cell Biology, 2006. 7(4): p. 265-275. 4. Bao, G. and S. Suresh, Cell and molecular mechanics of biological materials. Nature Materials, 2003. 2(11): p. 715-725. 5. Kitano, H., Computational systems biology. Nature, 2002. 420(6912): p. 206-210. 6. Lecuit, T. and P.F. Lenne, Cell surface mechanics and the control of cell shape, tissue patterns and morphogenesis. Nature Reviews Molecular Cell Biology, 2007. 8(8): p. 633-644. 7. Conn, P.M., Imaging and spectroscopic Analysis of Living Cells: Optical and Spectroscopic Techniques. Imaging and Spectroscopic Analysis of Living Cells: Optical and Spectroscopic Techniques, 2012. 504: p. 1-506. 8. Jaiswal, J.K., et al., Use of quantum dots for live cell imaging. Nature Methods, 2004. 1(1): p. 73-78. 9. Stephens, D.J. and V.J. Allan, Light microscopy techniques for live cell Imaging. Science, 2003. 300(5616): p. 82-86. 10. Marquet, P., C. Depeursinge, and P.J. Magistretti, Exploring Neural Cell Dynamics with Digital Holographic Microscopy. Annual Review of Biomedical Engineering, Vol 15, 2013. 15: p. 407-431. 11. Sung, Y.J., et al., Optical diffraction tomography for high resolution live cell imaging. Optics Express, 2009. 17(1): p. 266-277. 12. Marquet, P., et al., Digital holographic microscopy: a noninvasive contrast imaging technique allowing quantitative visualization of living cells with subwavelength axial accuracy. Optics Letters, 2005. 30(5): p. 468-470. 13. Frigault, M.M., et al., Live-cell microscopy - tips and tools. J Cell Sci, 2009. 122(6): p. 753-767. 14. Valiahdi, S.M., et al., Influence of extracellular pH on the cytotoxicity, cellular accumulation, and DNA interaction of novel pH-sensitive 2-aminoalcoholatoplatinum(II) complexes. Journal of Biological Inorganic Chemistry, 2013. 18(2): p. 249-260. 15. Busa, W.B., Mechanisms and Consequences of Ph-Mediated Cell Regulation. Annual Review of Physiology, 1986. 48: p. 389-402. 16. Griffiths, L., et al., The influence of oxygen tension and pH on the expression of platelet- derived endothelial cell growth factor thymidine phosphorylase in human breast tumor cells grown in vitro and in vivo. Cancer Research, 1997. 57(4): p. 570-572. 17. Davidson, J.F., et al., Oxidative stress is involved in heat-induced cell death in Saccharomyces cerevisiae. Proceedings of the National Academy of Sciences of the United States of America, 1996. 93(10): p. 5116-5121. 18. Pasantes-Morales, H., et al., Mechanisms of cell volume regulation in hypo-osmolality. American Journal of Medicine, 2006. 119: p. S4-S11. 19. Squirrell, J.M., et al., Long-term two-photon fluorescence imaging of mammalian embryos without compromising viability. Nature Biotechnology, 1999. 17(8): p. 763-767.

155

20. Sekar, R.B. and A. Periasamy, Fluorescence resonance energy transfer (FRET) microscopy imaging of live cell protein localizations. Journal of Cell Biology, 2003. 160(5): p. 629-633. 21. Michalet, X., et al., Quantum dots for live cells, in vivo imaging, and diagnostics. Science, 2005. 307(5709): p. 538-544. 22. Lippincott-Schwartz, J. and G.H. Patterson, Development and use of fluorescent protein markers in living cells. Science, 2003. 300(5616): p. 87-91. 23. Hoebe, R.A., et al., Controlled light-exposure microscopy reduces photobleaching and phototoxicity in fluorescence live-cell imaging. Nature Biotechnology, 2007. 25(2): p. 249-253. 24. Shaner, N.C., P.A. Steinbach, and R.Y. Tsien, A guide to choosing fluorescent proteins. Nature Methods, 2005. 2(12): p. 905-909. 25. Barty, A., et al., Quantitative optical phase microscopy. Optics Letters, 1998. 23(11): p. 817-819. 26. Liang, C.C., A.Y. Park, and J.L. Guan, In vitro scratch assay: a convenient and inexpensive method for analysis of cell migration in vitro. Nature Protocols, 2007. 2(2): p. 329- 333. 27. Allen, R.D., N.S. Allen, and J.L. Travis, Video-Enhanced Contrast, Differential Interference Contrast (Avec-Dic) Microscopy - a New Method Capable of Analyzing Microtubule-Related Motility in the Reticulopodial Network of Allogromia Laticollaris. Cell Motility and the Cytoskeleton, 1981. 1(3): p. 291-302. 28. Dodt, H.U. and W. Zieglgansberger, Visualizing Unstained Neurons in Living Brain- Slices by Infrared Dic-Videomicroscopy. Brain Research, 1990. 537(1-2): p. 333-336. 29. Shepherd, J.N.H., et al., 3D Microperiodic Hydrogel Scaffolds for Robust Neuronal Cultures. Advanced Functional Materials, 2011. 21(1): p. 47-54. 30. Tanaami, T., et al., High-speed 1-frame/ms scanning confocal microscope with a microlens and Nipkow disks. Applied Optics, 2002. 41(22): p. 4704-4708. 31. Nakano, A., Spinning-disk confocal microscopy - A cutting-edge tool for imaging of membrane traffic. Cell Structure and Function, 2002. 27(5): p. 349-355. 32. Jaiswal, J.K., et al., Long-term multiple color imaging of live cells using quantum dot bioconjugates. Nature Biotechnology, 2003. 21(1): p. 47-51. 33. Pawley, J., Fundamental Limits in Confocal Microscopy. Handbook of Biological Confocal Microscopy, Revised Edition, 1989: p. 15-26. 34. Wang, Z., et al., Spatial light interference microscopy (SLIM). Optics Express, 2011. 19(2): p. 1016-1026. 35. Rotsch, C. and M. Radmacher, Drug-induced changes of cytoskeletal structure and mechanics in fibroblasts: An atomic force microscopy study. Biophysical Journal, 2000. 78(1): p. 520-535. 36. Radmacher, M., et al., Measuring the viscoelastic properties of human platelets with the atomic force microscope. Biophysical Journal, 1996. 70(1): p. 556-567. 37. Curtis, M.W. and B. Russell, Micromechanical regulation in cardiac myocytes and fibroblasts: implications for tissue remodeling. Pflugers Archiv-European Journal of Physiology, 2011. 462(1): p. 105-117. 38. Radmacher, M., et al., From Molecules to Cells - Imaging Soft Samples with the Atomic Force Microscope. Science, 1992. 257(5078): p. 1900-1905. 39. Heinisch, J.J., et al., Atomic force microscopy - looking at mechanosensors on the cell surface. J Cell Sci, 2012. 125(18): p. 4189-4195.

156

40. Le, A.P., et al., Quantitative Reflection Imaging of Fixed Aplysia californica Pedal Ganglion Neurons on Nanostructured Plasmonic Crystals. Journal of Physical Chemistry B, 2013. 117(42): p. 13069-13081. 41. Stewart, M.E., et al., Quantitative multispectral biosensing and 1D imaging using quasi- 3D plasmonic crystals. Proceedings of the National Academy of Sciences of the United States of America, 2006. 103(46): p. 17143-17148. 42. Yao, J.M., et al., Seeing molecules by eye: Surface plasmon resonance imaging at visible wavelengths with high spatial resolution and submonolayer sensitivity. Angewandte Chemie- International Edition, 2008. 47(27): p. 5013-5017. 43. Stewart, M.E., et al., Nanostructured plasmonic sensors. Chemical Reviews, 2008. 108(2): p. 494-521. 44. Stewart, M.E., et al., Multispectral Thin Film Biosensing and Quantitative Imaging Using 3D Plasmonic Crystals. Analytical Chemistry, 2009. 81(15): p. 5980-5989. 45. Yao, J.M., et al., Functional Nanostructured Plasmonic Materials. Advanced Materials, 2010. 22(10): p. 1102-1110. 46. Ye, X.Y., et al., Production of Nitric Oxide within the Aplysia californica Nervous System. Acs Chemical Neuroscience, 2010. 1(3): p. 182-193. 47. Wang, L.P., et al., A Novel Pyridoxal 5 '-Phosphate-dependent Amino Acid Racemase in the Aplysia californica Central Nervous System. Journal of Biological Chemistry, 2011. 286(15): p. 13765-13774. 48. Caruso, F., et al., Ultrathin multilayer polyelectrolyte films on gold: Construction and thickness determination .1. Langmuir, 1997. 13(13): p. 3422-3426. 49. Richert, L., et al., pH dependent growth of poly(L-lysine)/poly(L-glutamic) acid multilayer films and their cell adhesion properties. Surface Science, 2004. 570(1-2): p. 13-29. 50. Curtis, H. and N.S. Barnes, Biology. 5th ed1989: W. H. Freeman. 51. Hine, R., The Facts on File Dictionary of Biology. 3rd ed. Facts on File1999: Checkmark Books. 52. Giebel, K.F., et al., Imaging of cell/substrate contacts of living cells with surface plasmon resonance microscopy. Biophysical Journal, 1999. 76(1): p. 509-516. 53. Wang, D., R. Tamburo, and G. Stetten, Gaussian Curve Fitter for Boundary Parameterization. MICCAI 2005 Workshop on Open-Source Software issue of The Insight Journal, 2005. 54. Oyama, Y., et al., Influences of Trypsin and Collagenase on Acetylcholine Responses of Physically Isolated Single Neurons of Aplysia-Californica. Cellular and Molecular Neurobiology, 1990. 10(2): p. 193-205. 55. Zhao, B., et al., Cell detachment activates the Hippo pathway via cytoskeleton reorganization to induce anoikis. Genes & Development, 2012. 26(1): p. 54-68. 56. Pasparakis, G., et al., Laser-Induced Cell Detachment and Patterning with Photodegradable Polymer Substrates. Angewandte Chemie-International Edition, 2011. 50(18): p. 4142-4145.

157

4.8 Figures

Figure 4.1 Digital picture of a home-built liquid cell for ellipsometry.

158

Figure 4.2 Schematic illustration of plasmonic reflection imaging system setup with laboratory microscope, bandpass filters, CCD camera, and full-3D plasmonic crystal. (Inset) A close-up illustration of cross-section of full-3D plasmonic crystal.

159

Figure 4.3 (a) Schematic illustration of cross section of full-3D plasmonic crystal used for FDTD modeling. The full-3D plasmonic crystal with the periodicity of hole of 740 nm, hole depth (Hdepth) of 280 nm, and diameter of hole (Dhole) of 544 nm are used for all FDTD calculation. For simplicity, we assumed that metal films are uniformly distributed on NOA nanoholes with a right angle edge; the thickness of the Au layer on top surface of the crystal (Ttop) and on the sidewall (Tside) and bottom (Tbottom) of the nanoholes are 44, 16, 16 nm, respectively. (b) Experimental transmission spectra (black) and FDTD calculated transmission spectra (red) of full-3D plasmonic crystal.

160

Figure 4.4 Schemes of polyelectrolyte film pattering process using permanent maker: (a) MHDA monolayer is formed on gold-coated silicon wafer by immersing the substrate in ethanolic solution of MHDA (2 mM). (b) Some lines are draw with permanent maker on a substrate to be used as mask for polyelectrolyte deposition. (c) Polyelectrolyte assemblies are grown on the substrate through layer-by-layer method. (d) Permanent marker mask is removed by rinsing with acetone.

161

Figure 4.5 (a) Three ellipsometric spectra of gold-coated silicone piece with MHDA monolayer; measured before polyelectrolyte deposition, after 5 layers of PAH/PSS deposition, and followed by immersing acetone. Ellipsometric spectra of gold-coated silicon piece immersed in acetone for 2 min is obscured by ellipsometric data obtained after 5 layers of PAH/PSS deposition. (b) 3D AFM image of the step edge of a polyelectrolyte pattern made by the deposition of 18 layers of PAH/PSS.

162

Figure 4.6 The thickness values of polyelectrolyte assemblies measured at increasing numbers of polyelectrolyte (PEL) layer deposition using AFM and ellipsometer.

163

Figure 4.7 FDTD simulated reflection spectra for pairs of optically equivalent polyelectrolyte and index-corrected material for the live cell conformally covering a full-3D plasmonic crystal: (a) 12 nm polyelectrolyte and 20 nm index-corrected material; (b) 24 nm polyelectrolyte and 44 nm index-corrected material; (c) 40 nm polyelectrolyte and 72 nm index-corrected material.

164

Figure 4.8 Corresponding pairs of thicknesses for polyelectrolyte films (n = 1.48) and index corrected mass coverage (θ) of live cell (n = 1.42).

165

Figure 4.9 FDTD simulated reflection spectra for different thickness of ASW gap (0, 12, and 24 nm of ASW gap) between plasmonic crystal and (a) 20 nm of index-corrected material and (b) 40 nm of index-corrected material.

166

Figure 4.10 (a) Reflection image (using a 500-550 nm bandpass filter) of a polyelectrolyte layer- by-layer assembly (PEL) and photocured polyurethane film (NOA) on a plasmonic crystal used for the reflection contrast calibration. Average pixel intensities were calculated in the regions marked with red box (NOA region) and blue box (PEL region), and used to determine the reflection contrast calibration. The scale bar corresponds to 100 μm. (b) Step-edge profile of normalized reflection contrast (blue circles) along the yellow arrow in Figure 4.10a and fitted red curve with a Gaussian width of 0.84 μm.

167

Figure 4.11 Normalized reflection contrast (NRC) as a function of index corrected thickness (θ) using different bandpass filters: 500-550 nm (black squares), 570-1000 nm (red circles), 570-600 nm (blue triangles), and 600-700 nm (pink inverted triangles). The error bars are obscured by the data markers themselves as a result of the large number of pixels (more than 30,000) used to determine the value.

168

Figure 4.12 (a) FDTD calculated reflection spectra for varying thickness of index corrected material on a nanostructured plasmonic crystal. (b) FDTD calculated reflection spectra for varying thicknesses of index-corrected material on flat gold film.

169

Figure 4.13 (a, b) Reflection images of a live Aplysia neuron cultured on a plasmonic crystal surface acquired using (a) 500-550 nm bandpass filter and (b) 570-600 nm bandpass filter. The scale bar in (a) and (b) corresponds to 100 μm. (c, d) Mass coverage transformed images generated by applying normalized reflection contrast calibrations to images of a live Aplysia neuron cultured on a plasmonic crystal acquired using (c) 500-550 nm bandpass filter and (d) 570-600 nm bandpass filter. The z-scale for image is restricted to the plasmonic sensing volume (0-80 nm). (e, f) Height profiles (along the white arrow) of the quantitative reflection image acquired using (e) 500-550 nm bandpass filter and (f) 570-600 nm bandpass filter.

170

Figure 4.14 Raw reflection image of live Aplysia neuron cells cultured on a plasmonic crystal acquired using (a) a 500-550 nm bandpass filter and (b) a 570-600 nm bandpass filter. The scale bar corresponds to 100 μm. The regions in the red boxes are magnified and present in Figure 4.13.

171

Figure 4.15 (a) AFM image of a fixed Aplysia neuron after FFT filtering to remove plasmonic nanostructure contribution. Imaged neuron cell is identical with one presented in Figure 4.13. (b) AFM height profile (fixed cell) along the white arrow drawn on Figure 4.15a and estimated thickness of live cell by consideration of volume changes resulting from water removal.

172

Figure 4.16 Mass coverage transformed images of the live Aplysia neuron cell acquired for time- lapse analysis of cell detachment induced by trypsinization. The scale bars correspond to 100 μm.

173

Figure 4.17 Mass coverage transformed images of a live Aplysia neuron cell acquired (a) before and (b) 60 min after the injection of trypsin. The scale bar corresponds to 100 μm. (c) Line profiles of index corrected mass coverage along the white arrows in Figure 4.17a (black line) and Figure 4.17b (red line). (d) Three reflection spectra of plasmonic crystals covered with 36 of index corrected mass coverage, 60 of index corrected mass coverage, and 160 nm ASW gap between plasmonic crystal and 60 of index corrected mass coverage.

174

CHAPTER 5

FUTURE RESEARCH DIRECTIONS FOR QUANTITATIVE IMAGING APPLICATIONS USING NANOSTRUCTURED PLASMONIC CRYSTALS

5.1 Sensitivity Improvement of Quantitative Plasmonic Imaging Technique for Live

Cell

As described in Chapter 3, the contrast of plasmonic reflection images could be enhanced by combining images in a multispectral form, which makes the imaging technique more sensible for small features.[1] In addition to the resultant imaging processing, improving the sensitivity of the plasmonic crystal itself can be another efficient approach to obtain a more accurate quantitative result. The plasmonic features generated on a plasmonic crystal are completely influenced by any slight variation of nanostructure such as size, shape, composition, and distribution.[2-5] This flexibility provides versatile application potentials of plasmonic crystals through application-specific performance optimization. To find the optimal design rules of a plasmonic crystal for different applications, FDTD computation is a powerful tool because optical behavior of a plasmonic crystal can be theoretically predicted without intensive physical experiments.[6] Thus, the performance of quantitative reflection image can be improved easily by engineering nanostructures on plasmonic crystals. For example, we demonstrated that the nanohole array with 580 nm of periodicity and 300 nm of diameter behaves more sensitively in the visual light range compared to one with 740 nm of periodicity and 440 nm of diameter

(described in Chapter 2). It is therefore theoretically predictable that a plasmonic crystal with smaller periodicity responds more sensitively to thin film thickness changes than with larger periodicity.

175

5.2 Quantitative Imaging Technique Using Functionalized Plasmonic Crystals

The cell imaging technique described in this dissertation was employed to measure the thickness of a cell or distance between a cell and a substrate. This SPR-based technique, however, can be further implemented for detecting and quantifying specific chemical/biological molecules by functionalizing the surface of plasmonic crystals.

Cells produce signals by protein secretion to communicate to other cells in the vicinity.

The ability to map the spatiotemporal nature of individual cell secretions is thus fundamental to understanding a cell’s vital functions including cell migration and proliferation.[7, 8] In our earlier works, plasmonic crystals were employed as a platform of the quantitative sensor for detecting IgG antibody/antigen pair.[9] This previous result demonstrated that antigen functionalized plasmonic crystals could be utilized as an imaging system for quantitative determination of antibody concentrations in the vicinity of a cell, as well as for estimation of the antibody diffusion constant in the extracellular media.[10]

Hydrogels undergo swelling or deswelling in response to pH changes or stimulations with specific biomolecules such as glucose or protein.[11] A plasmonic crystal functionalized with a pH sensitive hydrogel was previously validated as a highly sensitive pH sensor capable of detecting even 0.1 pH unit changes.[12, 13] Neuronal activity shifts extracellular or intracellular pH to modulate neuronal metabolism.[14, 15] These pH changes within a cell’s microenvironment lead to changes in the volume of the pH-sensitive hydrogel, which can be measured by optical response changes of plasmonic crystals. Therefore, functionalized plasmonic crystals are remarkably promising as a platform for monitoring cell dynamics in real- time.

176

5.3 Characterizing Thickness and Refractive Index Using Two-Wavelength Analysis

In this dissertation, we suggest a SPR-based optical imaging technique that is capable of quantifying the thickness of transparent thin films with well-known refractive index. Although this method is powerful to determine the thickness of an ultrathin film in the range of less than

100 nm, it is insufficient to analyze quantitatively thicknesses without prior knowledge of the thin film.[1] Other techniques commonly used for thin film characterization such as ellipsometry, quartz crystal microbalance (QCM), and atomic force microscopy (AFM) also have limitations to simultaneously obtain both thickness and refractive index of a thin film. Ellipsometric response is determined by both dielectric constant and thickness of a thin film, which makes it difficult to separate the two parameters (thickness and dielectric constant) from the resultant data.[16, 17] QCM and AFM are alternative tools to accurately measure the thickness of a thin film, but those are not applicable to characterization of optical properties of a thin film.[18, 19]

Some previous works reported that multispectral or multi-medium analysis using prism- based surface plasmon resonance spectroscopy allows simultaneous determination of both thickness and refractive index of transparent thin films.[20, 21] Briefly, this technique employs multiple plasmonic data obtained from the different conditions (wavelength or medium) to calculate thickness versus refractive index (d vs. n) continuum solutions for each condition using theoretical modeling. The thickness and refractive index of a thin film corresponds to the common solution of these multiple d vs. n continuum solutions. Because the optical behavior of plasmonic crystals is wavelength-dependent,[1] the multispectral analysis method can be applied to plasmonic reflection imaging system. The wavelength-dependent d vs. n continuum solutions for certain reflection contrast can be calibrated using FDTD calculations, which is then applied to simultaneously determine the thickness and refractive index of the thin film from the

177 reflection contrast values acquired using a different wavelength. Although prism-based and grating-based SPR systems are founded on the identical underlying physics, plasmonic crystals have more interesting potential as a platform for in situ visualizing technique without a demanding optical setup.

5.4 Final Remarks

The work discussed in this dissertation has demonstrated analytical applications using nanostructured plasmonic crystals for chemical/biological sensing and imaging. Because the optical behavior of plasmonic crystals is simply tunable by adjusting size, shape, composition and distribution of nanostructures, plasmonic crystals can be utilized as a versatile and powerful platform for surface-sensitive sensing and field-enhanced applications. In addition, the highly- uniform nanohole arrays facilitate the construction of a label-free and steadily performed sensing tool at low cost. Due to these advantages, plasmonic crystals have an excellent potential for numerous analytical applications

In this chapter, some future research ideas are proposed in an attempt to extend the range of applications or to enhance the performance of the interesting findings discussed in this entire dissertation. Studies based on these research directions using plasmonic crystals also can serve as a foundation for further inventions of chemical sensing and imaging techniques. Due to their sensitivity, application-flexibility, and cost efficiency, analytical applications using plasmonic crystals are anticipated to be widely used in real-life, beyond limited laboratory use.

178

5.5 References

1. Le, A.P., et al., Quantitative Reflection Imaging of Fixed Aplysia californica Pedal Ganglion Neurons on Nanostructured Plasmonic Crystals. Journal of Physical Chemistry B, 2013. 117(42): p. 13069-13081. 2. Kelly, K.L., et al., The optical properties of metal nanoparticles: The influence of size, shape, and dielectric environment. Journal of Physical Chemistry B, 2003. 107(3): p. 668-677. 3. Hutter, E. and J.H. Fendler, Exploitation of localized surface plasmon resonance. Advanced Materials, 2004. 16(19): p. 1685-1706. 4. Homola, J., I. Koudela, and S.S. Yee, Surface plasmon resonance sensors based on diffraction gratings and prism couplers: sensitivity comparison. Sensors and Actuators B-Chemical, 1999. 54(1-2): p. 16-24. 5. Ghosh, S.K., et al., Solvent and ligand effects on the localized surface plasmon resonance (LSPR) of gold colloids. Journal of Physical Chemistry B, 2004. 108(37): p. 13963-13971. 6. Chang, S.H., S.K. Gray, and G.C. Schatz, Surface plasmon generation and light transmission by isolated nanoholes and arrays of nanoholes in thin metal films. Opt Express, 2005. 13(8): p. 3150-3165. 7. Friedl, P. and D. Gilmour, Collective cell migration in morphogenesis, regeneration and cancer. Nature Reviews Molecular Cell Biology, 2009. 10(7): p. 445-457. 8. Heldin, C.H. and B. Westermark, Mechanism of action and in vivo role of platelet- derived growth factor. Physiological Reviews, 1999. 79(4): p. 1283-1316. 9. Stewart, M.E., et al., Multispectral thin film biosensing and quantitative imaging using 3D plasmonic crystals. Anal Chem, 2009. 81(15): p. 5980-5989. 10. Raphael, M.P., et al., Quantitative Imaging of Protein Secretions from Single Cells in Real Time. Biophys J, 2013. 105(3): p. 602-608. 11. Miyata, T., T. Uragami, and K. Nakamae, Biomolecule-sensitive hydrogels. Advanced Drug Delivery Reviews, 2002. 54(1): p. 79-98. 12. Mack, N.H., et al., Optical transduction of chemical forces. Nano Lett, 2007. 7(3): p. 733-737. 13. Marcombe, R., et al., A theory of constrained swelling of a pH-sensitive hydrogel. Soft Matter, 2010. 6(4): p. 784-793. 14. Blaustein, M.P., J.P.Y. Kao, and D.R. Matteson, Cellular physiology and neurophysiology. 2nd ed. Mosby physiology monograph series2012, Philadelphia, PA: Elsevier/Mosby. xx, 337 p. 15. Carlin, K.P., Modulation of calcium currents in mouse ventral horn neurons by extracellular pH. European Journal of Neuroscience, 2005. 22(10): p. 2655-2660. 16. Rodenhausen, K.B., et al., Combined optical and acoustical method for determination of thickness and porosity of transparent organic layers below the ultra-thin film limit. Review of Scientific Instruments, 2011. 82(10). 17. Azzam, R.M.A. and N.M. Bashara, Ellipsometry and polarized light1977: North-Holland Pub. Co. xvii, 529 p. 18. Buttry, D.A. and M.D. Ward, Measurement of Interfacial Processes at Electrode Surfaces with the Electrochemical Quartz Crystal Microbalance. Chem Rev, 1992. 92(6): p. 1355-1379. 19. Hansma, H.G. and L. Pietrasanta, Atomic force microscopy and other scanning probe microscopies. Current Opinion in Chemical Biology, 1998. 2(5): p. 579-584.

179

20. Peterlinz, K.A. and R. Georgiadis, Two-color approach for determination of thickness and dielectric constant of thin films using surface plasmon resonance spectroscopy. Optics Communications, 1996. 130(4-6): p. 260-266. 21. Granqvist, N., et al., Characterizing Ultrathin and Thick Organic Layers by Surface Plasmon Resonance Three-Wavelength and Waveguide Mode Analysis. Langmuir, 2013. 29(27): p. 8561- 8571.

180