National Academies Press: OpenBook

Nanophotonics: Accessibility and Applicability (2008)

Chapter: 3 Enabling Technologies

« Previous: 2 Nanoscale Phenomena Underpinning Nanophotonics
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 83
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 84
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 85
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 86
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 87
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 88
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 89
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 90
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 91
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 92
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 93
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 94
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 95
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 96
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 97
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 98
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 99
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 100
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 101
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 102
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 103
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 104
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 105
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 106
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 107
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 108
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 109
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 110
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 111
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 112
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 113
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 114
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 115
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 116
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 117
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 118
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 119
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 120
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 121
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 122
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 123
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 124
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 125
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 126
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 127
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 128
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 129
Suggested Citation:"3 Enabling Technologies." National Research Council. 2008. Nanophotonics: Accessibility and Applicability. Washington, DC: The National Academies Press. doi: 10.17226/11907.
×
Page 130

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

3 Enabling Technologies The development of enabling technologies, along with significant infrastructure, will be essential for the implementation of commercial and military applications of nanophotonics. This chapter describes essential enabling technologies, including the synthesis, growth, and fabrication of nanomaterials and nanostructures; modeling and simulation; characterization techniques for nanophotonics; and the pack- aging and integration of nanophotonics devices. The development of these enabling technologies in a country is often an important indicator of the state of maturity of nanophotonics in that country. Realizing Hierarchical Synthesis, Growth, and Fabrication Structures at the Nanoscale Introduction Traditionally, synthesis, growth, and fabrication have been separately identified stages in the devel- opment of functional devices. An example is a semiconductor laser: the synthesis is in developing the nearly defect-free substrate (gallium arsenide [GaAs], for example) and in developing the feed materials for the epitaxial crystal overgrowth. The growth stage as related to a semiconductor laser is the epitaxial formation of an optical cavity and gain structure for the appropriate confinement of both the photons and the electronic carriers using molecular-beam epitaxy or metal-organic chemical vapor deposition (MOCVD), typically on a wafer scale. The fabrication stage uses processes adapted largely from inte- grated circuit manufacturing to batch fabricate diode lasers with contacts for electrical input and optical output as required for the device application. In the nanoscale era, these distinctions among synthesis, growth, and fabrication are blurring. The steps outlined above are being employed in mix-and-match ways to produce novel functional nanostructured materials and to arrange them with the necessary hierarchical organization to produce new functionalities. This subsection describes the traditional categories of synthesis, growth, and fabrication, and then comments on the blurring of the boundaries among these stages. As the terms are used here, synthesis includes not only the starting materials, but also the formation of isolated nanostructures, such as metal 83

84 Nanophotonics and dielectric nanospheres and quantum dots, and fabricated multilayer combinations. Additionally, the chemical synthesis of starting materials such as copolymers and liquid crystals is included. Growth as used here is specifically limited to the epitaxial growth of semiconductor materials; it also includes self-assembled arrays via nanoparticle formation, such as Stranski-Krastanov formation of quantum dots by the interplay between strain and surface tension. Fabrication refers to the creation of ensembles of nanostructures—for example, photonic crystals composed of nanoparticles. Traditional top-down processing derived from the integrated circuit industry is reaching to scales of direct relevance to nanophotonics and beyond and certainly will be an important component of any nanophotonics fabrication suite. This is so in part as a result of the strong ability of traditional top-down processing to engineer hierarchical structures incorporating multiple, disparate length scales and the proven mass production capabilities of batch wafer processing. Self-assembly is bottom-up processing; a simple example is the assembly of colloidal nanoparticles into photonic crystal arrangements. A related approach is the use of multiphase systems, such as block copolymers along with surfactant-covered nanoparticles, to spontaneously form complex patterns driven by an external forcing function such as evaporation. Increasingly, techniques are being developed that combine top-down and bottom-up approaches and blur the distinctions of the categories described above. One example is nanoscale crystal growth, in which a pattern is defined by fabrication and a subsequent growth process results in an array of nanoscale semiconductor structures. The fabrication can occur either by self-assembly (as, for example, porous anodization of a continuous metal film) or by traditional lithographic pattern definition and etching. As opposed to the subtractive etching of a large-area semiconductor film (the growth described above), this sequence allows much more flexibility and often produces materials with fewer defects and improved functionality. Generically, this combination is referred to as directed self-assembly. Figure 3-1 shows a lithographic template that serves to guide the self-organizing crystals, block polymers, or colloids to pack and orient into a predetermined pattern, thus bringing the ultrasmall length scales achievable by molecular self-organization into play with the somewhat larger engineered structures to create hierarchical devices. Synthesis Nanoparticles The need for improved optical materials has driven researchers in chemistry, materials science, and chemical engineering to create new synthetic pathways to afford better control over the composition, size, and shape of nanoparticles in order to produce new types of supramolecular building blocks. As mentioned in Chapter 2, metallic nanoparticles, such as gold nanostars, can be useful for localized surface plasmon resonance. Thus, it is important to control not only particle size but particle shape. Moreover, it is absolutely critical to have monodispersity in size and shape. Researchers have found that absorbed surfactants can influence the relative growth rate of various facets (surfactant binding and size-dependent facet surface energies) resulting in the growth of complex- shaped nanoparticles. For example, Alivisatos and colleagues found that they could create tetrapod gold (Au) or cadium selenium (CdSe) nanoparticles. The Au particles are composed of a set of (111) twin Block copolymers are made up of two or more homopolymer subunits linked by covalent bonds. Block copolymers with two (or three) distinct blocks are called diblock (or triblock) copolymers.

ENABLING TECHNOLOGIES 85 FIGURE 3-1  A top-down lithographically defined template can guide the bottom-up self-assembly of materials, creating a system with several length scales: a structural hierarchy. SOURCE: Cheng et al. (2006b). Copyright 2006 Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. 3-1 Au crystals (Manna et al., 2002). Additionally, hollow nanoparticles can be produced by the oxidation of metallic nanocrystals and directional diffusion of species (the Kirkendall effect) (Yin et al., 2004). Traditional solution synthesis has recently been extended to microfluidic reactors, in which on-chip in situ monitoring of the growing entities and the ability to add chemical feed streams and to adjust residence times provide opportunities for dynamic adjustment and tuning of the targeted materials. An example is shown in Figure 3-2, where an overcoating of zinc sulfide (ZnS) onto CdSe quantum dots (QDs) increases the quantum yield from ~10 percent to ~40 percent along the reactor pathway. The meandering synthesis channel is fabricated in silicon and is thermally isolated from the room-­temperature inlets and outlet. ZnS overcoating reagents enter through the side inlets to enable core-shell particle growth without secondary nucleation. Advances in microfluidic reactors will increasingly impact the ability to engineer many types of nanoparticles because, instead of batch reactions with a Schlenk line apparatus with multiple step- f ­ unction injections of reactants, one can tailor concentration-time schedules and use gradients and can separate and remix-recycle, impinge, and sinter or stack the growing targets to achieve desired structures and properties. Currently most particles are homogeneous or core-shell and either spherical

86 Nanophotonics FIGURE 3-2  Microfluidic reactor for the synthesis of zinc sulfide (ZnS) overcoated cadium selenide (CdSe) quantum dots. The brightness in the lower right image increases from bottom to top, indicative of the increased quantum efficiency as the ZnS overcoat layer forms on the CdSe dots. SOURCES: Yen (2006), Yen et al. (2005). 3-02 Reproduced with permission. or ­ cylindrical; however, multicomponent, complex-geometry particles are possible and can provide structural and functional anisotropy. Layered-Nanoparticle Fabrication Techniques Layered nanoparticles with a core-shell geometry in which the core and shell layers consist of either a dielectric material or a noble metal are of scientific interest, as the optical properties of these particles can be controlled by adjusting the thickness of each layer. The most common types of core-shell particles are nanoshells, which consist of a silica core coated with a thin layer of gold or silver (Oldenburg et al., 1998). The opposite particle consisting of a gold or silver colloid as the core particle coated with a silica shell of varying thickness can also be made (Liz-Marzan et al., 1996). The optical properties of these particles are controlled by the ratio of the core radius to the total particle radius. Similar to these, but nonspherical in shape, are spindle-shaped iron oxide particles coated with a thin layer of gold (Wang et al., 2006b). Core-shell semiconductor particles are also important—for example, a thin (<1 nm) ZnS outer shell passivates CdSe quantum dots. The silica core particles are typically synthesized using the Stöber process (Stöber et al., 1968). The Stöber method involves the base catalyzed hydrolysis and condensation of tetraethylorthosilicate (TEOS). Typically, TEOS and ammonium hydroxide (NH4OH) are mixed in varying ratios in ethanol to produce silica nanoparticles in the 80 nm to 500 nm diameter size regimes. After the silica particles are synthesized, the surface is functionalized with a silane, such as 3-aminopropyltriethoxysilane (APTES).

ENABLING TECHNOLOGIES 87 This chemical functionalization of the silica surface provides an amine moiety at the surface, which is used to attach ultrasmall gold colloid. Ultrasmall gold colloid particles of 1 nm to 3 nm in size are synthesized by reducing chloroauric acid using tetrakis (hydroxymethyl) phosphonium chloride (THPC) as a reducing agent. The ultrasmall gold attached to the silica surface act as nucleation sites for the electroless deposition of Au (or silver [Ag]) to form a complete shell on the silica core. The technique for depositing gold layers on spindle-shaped hematite particles is similar to depositing a gold layer on silica particles. The growth of silica layers on gold or silver colloidal particles follows an inverse process. A solution of gold particles with an average diameter of around 15 nm and 10 percent polydispersity is used to form the core of these layered particles. A freshly prepared aqueous solution of APTES (2.5 milliliters [ml], 1 millimeter [mm]) is added to 500 ml of the gold solution under vigorous magnetic stirring. The mixture of APTES and gold dispersion is allowed to stand for 15 minutes to ensure complete complexation of the amine groups with the gold surface. A solution of active silica is prepared by lowering the pH of a 0.54 weight percent sodium silicate solution to 10 to 11 by the progressive addition of cation exchange resin. Twenty milliliters of active silica are then added to 500 ml of the surface-modified gold solution, again under vigorous magnetic stirring. The resulting dispersion (pH ~8.5) is then allowed to stand for at least 1 day, so that the active silica polymerizes onto the gold particle surface. The silica shell thick- ness is about 2 nm to 4 nm after 24 hours. The particles can then be transferred into ethanol if further growth or chemical modification of the silica layer is intended. At this point, thicker silica shells can be grown using the Stöber method. Another class of these layered particles consists of monolayers of silica or polystyrene nanoparticles coated with a layer of metal (Dieringer et al., 2006). This geometry has also been used for ­plasmonic applications such as surface-enhanced spectroscopy. The fabrication process begins with the self- a ­ ssembly of monodisperse nanospheres to form a two-dimensional colloidal crystal. A substrate is pre- pared so that the nanospheres freely diffuse until they reach their lowest energy configuration. This is achieved by chemically modifying the nanosphere surface with a negative charge that is ­electrostatically repelled by a negatively charged substrate such as chemically treated glass. As the solvent (water) evapo- rates, capillary forces draw the nanospheres together, and they crystallize in a hexagonally close-packed pattern on the substrate. Following self-assembly of the nanosphere mask, a metal (typically silver) is deposited by physical vapor deposition from a collimated source normal to the substrate through the nanosphere mask to a controlled mass thickness. The resulting surface is referred to as a metal (e.g., Ag) film over nanosphere (AgFON) surface. Nanorods and Nanowires Nanorods and nanowires are another promising type of nanoparticle that can be made a variety of ways. Nanowires are quasi–one-dimensional single-crystalline structures and, in contrast to nanorods, they have a much larger length-to-radius aspect ratio, with diameters as small as a few nanometers and lengths up to several hundreds of micrometers. Sized controlled nanorods (Hu et al., 2001) can be made by thermal decomposition of organometallic precursors in a coordinating organic solvent. The ­discovery that various ligands and co-ligands can promote anisotropic growth has led researchers to create a ­number of nanorods that afford uniform size and shape and good electrooptical properties, controlled by the molar ratio of the components to the coordinating solvent as well as annealing time. Nanowires are typically grown using chemical vapor deposition (CVD) by means of a catalyst- mediated vapor-liquid-solid (VLS) mechanism. The growth is initiated by the dissolution of gaseous reactants in nanosized liquid particles, followed by nucleation and growth on the substrate. In this way,

88 Nanophotonics the nanowire diameter is predefined by the size of catalytic particles, length depends on the growth time, and often there is an epitaxial relationship between the nanowire and the substrates allowing control of the nanowire growth direction (Lu and Lieber, 2006). In addition, nanowires can be grown in place by a number of techniques, allowing an ensemble of wires to be grown all having the same crystal orienta- tion. One technique uses VLS growth with a low concentration of nickel nitrate catalyst on a sapphire substrate to produce dense arrays of gallium nitride (GaN) nanowires, well aligned to each other and having a vertical orientation (Wang et al., 2006a). Another extremely promising technique is that of using conventional MOCVD growth to produce uniformly oriented and ordered arrays of GaN nanowires by growing through a nanopatterned mask material of silicon nitride or silicon dioxide onto a GaN film. By using a “pulsed” growth mode in which precursor gases are introduced alternately in sequence, the diameter of the quantum wires can be made extremely uniform along their length (Hersee et al., 2006). Additional description of the latter technique is provided later in this chapter. Semiconductor nanowires can simultaneously act as interconnects and as active components and therefore have emerged as attractive building blocks for applications ranging from nanoelectronics to photonics or sensing. Through the rational growth of nanowire heterostructures with controllable dop- ing, the functionality of such devices has been enhanced, and nanowire-based light-emitting devices, high-performance field-effect transistors (FETs), and arrays of single-nanowire silicon FETs for sensing have been demonstrated (Xiang et al., 2006). Central to future applications will be the ability to assemble and position such nanoscale devices at different length scales, as well as the ability to address individual elements in high-density arrange- ments. For example, microfluidic flows can be used for aligning an array of nanowires in devices (see Figure 3-3), while for the manipulation of individual nanowires several techniques have been proposed: optical tweezers, or manipulation using electrical and magnetic fields. Organic Materials Organic nanophotonic materials are more difficult to control with the structural precision of inorganic materials, but chemists can create intricate molecules and supramolecular assemblies with extremely highly specific interactions with chemicals and biological agents. Also, organics have an enormous range of electrooptical properties and are flexible hosts for metallic and inorganic materials to create multi- functional devices. In addition, their huge tunability across the ultraviolet (UV) to infrared (IR) spectral regions by means of a host of stimuli (e.g., temperature [Wiersma and Cavalieri, 2001], mechanical stress, pH, ionic strength, specific binding to biological agents, response to electric fields, and so on) makes them excellent sensors (Gaillot et al., 2007; Moreira et al., 2004). Many industrial processes have been developed for the deposition of uniform organic films. Spin casting is the most familiar and produces films with nanometer-thickness resolution and control. Dry film-deposition processes have also been commercialized. For many device applications—for example, organic light-emitting diodes (OLEDs)—this approach has been very successful. Other device concepts require a precision placement and/or contacting of individual molecular moieties. A challenge for solution-based synthesis is to be able to place and to orient, contact, and interconnect these individual molecular moieties into nanoscale Certainly, chromonic liquid crystals are being used (two companies are developing this scheme) for the detection of ­specific biological agents. For more information, see (http://dept.kent.edu/biology/woolver.htm), (http://www.blackwell-synergy.com/ doi/pdf/10.1111/j.1472-765X.2006.01916.x). The current schemes are not PBG-based, but certainly can be and very likely will be.

ENABLING TECHNOLOGIES 89 FIGURE 3-3  Use of microfluidics to flow-align nanowires and form crossed arrays by using a poly­dimethylsiloxane (PDMS) stamp and successive directional flows. SOURCE: Lu and Lieber (2006). Reproduced with permission from Journal of Physics D: Applied Physics. devices. The use of nanoimprint lithography to create surface patterns with variable chemical affinity is emerging as a viable pathway to direct the assembly of organics. Co-assembly Instead of the self-organization of a single type of molecule during solvent evaporation, co-assembly encompasses the idea that multiple types of entities—various molecules, quantum dots, and so on—can be simultaneously organized from a liquid mixture as the temperature is lowered or as the solvent evapo- rates to form more-complex and often multiscale, multifunctional structures. An example is a two-size colloidal assembly to form Laves phases. The structure depends on the size and molar ratio of the small and large spheres and basically seeks to achieve maximum packing density. This idea has been used “Conventional nanoprint lithography often involves exposing the patterned polymer to high temperatures, UV exposure and etching processes. These processes result in a harsh environment that potentially degrades the electrical properties of the p ­ olymeric semiconductor. . . . Still other techniques use a surface-energy pattern on a substrate to pattern a polymer. C. R. Kagan et al. in Appl. Phys. Lett. 79 (21) 3536 (2001) describes patterning self-assembled monolayers using such a surface- energy pattern. Such patterns are typically generated using surface energy modulation. However, use of such a system in elec- tronic device fabrication is restricted to surfaces on which a self-assembled monolayer can be deposited (typically the noble metals such as gold or palladium). An additional coating step, typically accomplished through dip-coating the surface-energy pattern of the substrate over the entire substrate area is complex and slow, lowering throughput and yield” (see http://www.­ freepatentsonline.com/20060279018.html).

90 Nanophotonics to make novel superlattices of metallic nanoparticles, magnetic nanoparticles, and semiconductor nano­ particles (Shevchenko et al., 2006). Co-assembly of a block copolymer (BCP) and one or more types of nanoparticles can produce interesting hierarchically structured materials. The high-molecular-weight BCP acts as a nanostructured host matrix with a periodicity on the order of the wavelength of light, and the sequestration of various types of the much smaller nanoparticles (see the following subsection) can be used to locally alter the refractive index of the larger block structures. Synthesis of Block Polymers The synthesis of polymers is being extended to an ever-greater range of monomers to create unique diblock and multiblock polymers with both linear and highly branched architectures. The ability to com- bine two or more polymeric species into a block polymer presents the possibility for multi­functionality of properties owing to the self-assembly of the blocks into equilibrium one-, two-, and three-dimensional periodic nanophases and microphases having a variety of geometries. BCPs can also be used to ­sequester nanoparticles where, by appropriate choice of the surface ligand on the nanoparticles, the particles co-assemble with the BCP from solution into the targeted microdomains (Bockstaller et al., 2005). I ­ ncorporation of metallic and inorganic (semiconducting) particles can modify the dielectric properties of the assembly to provide, for example, increased dielectric contrast for photonic crystals. Liquid Crystals Liquid crystals spontaneously form modulated phases that are photonic band-gap materials and hence are interesting and useful for nanophotonics applications. The structures of the phases that are formed are determined by the properties of the molecules that are the building blocks of the liquid crys- tal. Rod-like, chiral low-molecular-weight molecules with aromatic cores and flexible alkyl chains form cholesteric and ferroelectric phases, which are periodic in one dimension. These are photonic band-gap materials, with a pitch that can be made to vary from tenths to tens of microns. Such structures can also be polymerized to make rigid cholesteric plastics, as well as weakly cross-linked to make mechanically deformable cholesteric “rubbers.” One application is the resulting tunable rubber distributed feedback (DFB) lasers shown in Figure 3-4. Liquid-crystal elastomers may also be used as photoactuators and nano-optomechanical system elements. Chiral molecules also exhibit the cholesteric “blue” phases, which are periodic in three-dimensions. Although in the past these have only existed in a very narrow temperature range, recent work (Coles et al., 2006) enabled the realization of materials that are stable over a range of 50ºC. Such robustness is expected to enable the synthesis of easily processible large-area flexible plastic photonic band-gap materials in the near future, although of lower index contrast (n2 – n1 < 0.3). Examples of such self-assembled structures are the helical phases of cholesteric liquid crystals, which are periodic in one-dimension and show strong Bragg reflection for a range of wavelengths and one polarization mode and mirrorless lasing at the band edge (see Figure 3-5). As shown in Figure 3-6, liquid crystals that are periodic in three-dimensions exist as well. These are the cholesteric and smectic blue phases, twisted grain-boundary phases, forming self-assembled three-dimensional photonic band-gap structures. Other relevant new materials developments are chromonic liquid crystals, in which the mesogenic building blocks are molecular aggregates without covalent bonding, and “banana” liquid crystals of achiral bent-core molecules that form periodic chiral phases.

ENABLING TECHNOLOGIES 91 a b FIGURE 3-4  Optically pumped mirrorless lasing by a dye-doped cholesteric liquid- crystal elastomer. The lasing wavelength is tuned by stretching the rubber laser (which results in compression of the one-dimensional Bragg reflector in the direction of the film thickness). (a) Demonstration of tunable lasing by deformation (DCM dye optically pumped at 532 nm); (b) changes in normal incidence reflectivity of cholesteric rubber structure due to biaxial deformation. SOURCE: Finkelmann et al. (2001). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. 3-4 part "a" is new higher res Nanoparticles with functionalized surfactants and tethers have the potential to self-assemble into a tremendous variety of structures. One example, predicted by numerical simulations and observed experimentally, is the spontaneous formation of flat sheets of cadmium telluride (CdTe) nanocrystals without the presence of templating surfaces. Figure 3-7 shows experimental demonstration of this type of structure formation, with CdTe nanocrystals forming into nanometer-thin sheets.

92 Nanophotonics a b 1.0 Reflectance 0.8 0.6 0.4 P 0.2 0.0 2 400 500 600 700 λ (nm) c 5 4 BLO61-E7:DCM(2%WT) Light Intensity (a.u.) 4 λex =355 nm. 3 3 2 1 0 2 1 0 500 600 700 Wavelength (nm) Reflection band Laser emission λem = 668 nm FIGURE 3-5  Self-assembled liquid crystals can form one-dimensional reflectors and exhibit band-edge lasing. SOURCES: (a, c) Palffy-Muhoray et al. (2006), reproduced with permission of SPIE; and (b) provided by com- mittee member Peter Palffy-Muhoray. 3-5

ENABLING TECHNOLOGIES 93 FIGURE 3-6  The blue phase is composed of an ordered array of linear defects that pack on a cubic lattice. SOURCE: Reprinted by permission from Cao et al. (2002). Copyright 2002 by Macmillan Publishers Ltd. 3-06 FIGURE 3-7  Self-assembly of cadmium telluride nanocrystals (left) into free-floating sheets (right). SOURCE: Tang et al. (2006). Reprinted with permission of AAAS. Recent research shows that materials, whose building blocks are not molecules, but nanoparticles, also show liquid-crystal phases. Examples are clays and mineral suspensions, such as gibbsite platelets and goethite nanorods, as well as synthesized semiconductor nanorods and viruses such as tobacco mosaic virus. The resulting anisotropy provides an additional degree of freedom in the design of the o ­ ptical properties of these metamaterials. Advances in the synthesis of molecules and of nanoparticles that aggregate to form hierarchical structures are certain to play a key role in the development of ­materials for emerging nanophotonics device applications.

94 Nanophotonics Self-Assembled responsive materials The self-assembly of block copolymers and liquid crystals and even liquid-crystalline block poly- mers can lead to a variety of photonic crystals whose properties can be externally tuned (Park et al., 2003). Block copolymers microphase separately on a length scale of the blocks driven by the competition between the tendency to reduce the interfacial free energy and to increase the conformational entropy of the chains. A variety of BCP-based photonic crystals have been made, including one-, two-, and three-dimensional periodic materials (Edrington et al., 2001; Urbas et al., 2002). Since these materials are “soft,” they can readily accommodate organic dyes and quantum dots, adding to their functionality, including the ability to form self-assembled lasers (Yoon et al., 2006). Essential to their responsive func- tionality, the strong variation of refractive index with temperature, the variation of the segment-segment interaction parameter with temperature and with solvent content, and the ability to swell the polymers with solvents and homopolymers as well as to significantly deform the periodicities give rise to a host of thermochromic, solvatochromic, and mechanochromic materials. The greatest tunable response arises from the coil-globule collapse transition where changes in volume can be up to 1,000 percent, affording highly responsive pH, ionic strength, electric field, and humidity sensors. A hydrophobic-hydrophilic BCP can have its stop band tuned from about 400 nm to 1,600 nm by varying salt content. Colloidal Synthesis The large effort at creating self-assembled photonic crystals to create a band gap in the visible spec- trum employing colloids demands the synthesis of highly monodisperse spheres at targeted diameters (on the order of the wavelength). Much effort has been made to create the inverse opal structure by directed assembly of spheres (typically polystyrene or silica) into a “face-centered cubic” structure, followed by infiltration of a high-dielectric material into the intersphere void space (e.g., by a sol-gel route to titanium dioxide [TiO2]), followed by etching of the spheres and the air/dielectric inverse opal structure with a band gap between the 8th and 9th bands for n2/n1~3. Recently, diamond- packed spheres and spheres in the pyrochlore structure have been proposed to produce much-improved low-order, easily opened band gaps due to a preferred structural geometry. The idea is to create a binary colloid analogue of the MgCu 2 Laves phase that has two sublattices with diamond-like symmetry (Shevchenko et al., 2006). The desire to achieve large-area periodic patterns at nanoscale dimensions has also led to a number of approaches that incorporate “self-assembly” strategies, generally using natural building blocks, such as nanocrystals, that can serve as the template for three-dimensional photonic crystal structures (Norris and Vlasov, 2001; Schroden et al., 2002; Xia et al., 2001). The challenges in those approaches lie in the achievement of perfect, long-range order, and at the same time being able to engineer important non- periodic features (such as waveguide “defects”). It is also not always possible with currently available materials to achieve a sufficient contrast in index using this approach. Epitaxial Growth Much of the nanophotonic enterprise is dependent on epitaxial growth techniques, particularly for semiconductor-based devices. Epitaxial refers to the crystallographic deposition of atoms onto a regular crystal lattice substrate, so that the entire grown optoelectronic device is very precisely arranged. Nanophotonics is dependent on state-of-the-art crystal growth methods in two ways: (1) the quan- tum well structures must be precisely controlled in both material composition and thickness in order to make optoelectronic structures at specific wavelengths of light, and (2) the purity must be sufficiently

ENABLING TECHNOLOGIES 95 high (and material disorder sufficiently low) that electronic energy levels in reduced-dimensionality structures (quantum wells [QWs], quantum wires, and quantum dots) are not broadened beyond fairly specific requirements, typically determined by the particular application. Molecular-Beam Epitaxy One of the more sophisticated techniques for growing complex and extremely precise nanophotonic structures is molecular-beam epitaxy (MBE). Here, material is deposited with atomic precision using evaporated molecular beams from heated elemental sources, in an ultrahigh vacuum environment. The impurity background of the material can be made extremely low. The composition of the material is controlled by moving shutters over the elemental sources so as to block or unblock the beams. This technique enables subatomic-layer precision over the thickness of the deposited layers, such as quantum wells and barriers, with a ≤1 percent control over the material composition—for example, the amount of aluminum (Al) in an AlxGa1–x As layer. Many forefront nanophotonic devices require astonishing levels of precision and control over the growth. One example is quantum cascade lasers (QCLs) operating at terahertz frequencies. In these structures, the electron transitions between two quantum well-defined energy levels are only a few milli­ electronvolts (meV) apart in energy. Thus, the background disorder potential in the QWs (the result of ionized impurities, interface roughness, and dopants) must be well below 1 meV in order for the energy levels to be well defined. Further, change in the thickness of the QW by less than a single atomic layer (averaged over the surface area, of course) will result in an unacceptably large change in wavelength. In addition, not only must the QW and barrier thicknesses be controlled with submonolayer precision, but the same identical QCL “unit cell” must be repeated on the order of 200 times in a single QCL stack in order to achieve sufficient gain for lasing. Each of the 200 periods must be identical to the others within the same subatomic-layer precision. Because the total thickness of the QCL structure is on the order of 10 microns, the flux rate of material leaving the elemental sources actually changes during the growth of a single structure and must be compensated for (Williams et al., 2006). These types of challenges are perhaps responsible for the fact that groups at only three or four institutions in the world have succeeded in repeatedly growing terahertz QCLs. These include groups at Cambridge University’s Cavendish Laboratories, the Vienna University of Technology, the Swiss Federal Institute of Technology in Zurich, and the Department of Energy’s Sandia National Laboratories. Just recently, successful terahertz QCL growth has also been reported by the University of Massachusetts at Lowell and by Spire Corporation and Trion, Inc., in the United States. Metal-Organic Chemical Vapor Deposition Although molecular-beam epitaxy gives the greatest degree of precision and control over the growth of compound semiconductor structures, it suffers from a number of drawbacks. One of them is the slow growth rate; in order for the mechanical shutter operations to achieve subatomic-monolayer precision, the growth rate must be very slow, on the order of 1 angstrom (Å) per second. Further, MBE growth machines typically accommodate a relatively small number of substrate wafers, since beam flux diver- gence over the growth area must be minimized. These two factors, taken together with the ultrahigh vacuum requirements, conspire to make MBE growth very expensive. The other major technique for producing nanoscale semiconductor heterostructures is metal-organic chemical vapor deposition, also known as metal-organic-vapor-phase epitaxy (MOVPE). In this case, precursor gases consisting of metal-organic complexes (e.g., trimethylaluminum, trimethylgallium,

96 Nanophotonics arsine) flow through a reactor across a heated substrate wafer surface, at pressures of 10 torr to 100 torr. The gases decompose at the wafer surface to leave the elemental constituents of the semiconductor compound being grown. Because of the relatively high pressure and the fact that MOCVD growth typically operates in the laminar flow regime and is controlled by gas-flow valves, the ability to control composition and thick- ness of epitaxial layers is not quite as good as in MBE. Nonetheless, it continues to improve and is by far the current major means of producing optical components. Similarly, the background impurity levels of MOCVD-grown material are not as high as in MBE. These shortcomings, however, are compensated for by the fact that the growth rates of MOCVD are 10 to 100 times greater than in MBE. Furthermore, the laminar flow nature of MOCVD renders the reactors amenable to scalable designs with multiwafer platens. This enables the cost of the material to be substantially reduced, and it is for this reason that MOCVD is by far the dominant growth technique for the production of optoelectronic devices for both military and civilian applications. (Note, however, that MOCVD cannot yet produce the most demanding growth structures, such as terahertz QCLs.) Growth Challenges Continuing research needs and challenges in crystal growth will, if successfully addressed, expand the ability to create new nanophotonic structures with new levels of sophistication and precision. Key areas of need include (1) understanding the relationship between stress, nanoscale compositional struc- ture, and transport and optoelectronic properties; (2) better understanding of MOCVD growth chemistry, including nonlinear relationships between precursor flow rates and film composition, parasitic reactions, nanoscale precipitants, and metastable solid phases on reactor walls; and (3) new techniques for the in situ monitoring of growth conditions, surface structures, and chemical reactions. Examples of the l ­ atter include reflectometry to determine surface structure, reflectivity-corrected pyrometry for accurate surface temperature monitoring, in situ x-ray diffractometry, and stress-monitoring of thin films during deposition. International Semiconductor Crystal Growth Expertise Expertise in semiconductor crystal growth largely mimics the development of a sophisticated s ­ emiconductor-based photonics industry. Thus, the United States, Europe, and Japan and Korea are today in leadership positions. There are extensive development efforts in Taiwan, especially aimed at consumer products such as light-emitting diodes (LEDs), and this expertise is rapidly migrating to China. Overall, semiconductor growth is an expensive and complex process that requires a substantial infrastructure (raw materials, growth apparatus, ancillary equipment, and large, dedicated laboratory facilities) and an end user to justify the investments. Thus, development of these capabilities requires long lead times, and new activities of this type should be very apparent; it is unlikely that new participants in such activities would appear abruptly. Fabrication Planar Processing Approaches The integrated circuit industry has developed a broad suite of manufacturing tools that allow fabrica- tion on scales that are immediately relevant to nanophotonics. At the writing of this report, the industry

ENABLING TECHNOLOGIES 97 is just beginning the volume production of circuits incorporating 45 nm gate transistors, beyond the scales necessary for many nanophotonic applications, with promise of reaching scales of approximately 10 nm within the 15-year outlook of this report (Semiconductor Industry Association, 2005a). Industrial manufacturing processes encompass a suite of techniques both to define an image on a wafer and to trans- fer that image into hard materials (semiconductors, metals, and dielectrics). Emerging nano­photonics techniques extensively leverage these techniques, with much work focused on single or layered two- dimensional structures. Extensions to three-dimensional patterning such as multibeam interference are being explored specifically for photonic crystal applications. Optical Lithography Optical lithography is well established as the manufacturing technology of choice. With recent developments such as immersion lithography and double-exposure techniques, it now appears that the hegemony of optical lithography still has a significant run ahead of it, despite the continuing refrain that the end of optical lithography as we know it is virtually upon us. In this context, it is worth noting that predictions of the end of optical lithography have held steady at “two generations out” for roughly the past 35 years! Any discussion of tools for nanophotonics necessarily involves consideration of the associated costs—in resources and in time. Thus, for example, electronic (e)-beam lithography can produce struc- tures at scales smaller than are possible from optical lithography and with an almost complete pattern flexibility (within the limits imposed by proximity effects) and is a staple of nanoscience research. Nonetheless, many applications require large areas (many square centimeters of nanopatterned material), for which e-beam lithography is not a viable approach as a result of its serial point-by-point writing modality and the consequently long times involved in addressing the large number of pixels in a large- area nanoscale image. A linear pixel resolution of 20 nm corresponds to a density of 2.5 × 1011 cm–2 (and at a typical e-beam writing speed of 50 megahertz [MHz], writing a square centimeter takes about 10 hours!). This issue only gets worse as the patterning gets deeper into the nanoscale. As an aside, this is a powerful argument applying to any technique, such as the various approaches to maskless lithography, that requires the storage and transfer of information on an individual-pixel basis. The data transfer demands are very difficult, even with today’s advances in computing and com- munications technologies. Optical lithography, in contrast, is a parallel writing technique. Traditional optical lithography uses a mask-based approach along with optical reduction to ameliorate the demands on both the mask fab- rication and on the optical system. Once the mask is fabricated, all of the information on the mask is transferred onto the wafer in the lithography step. The economics of the integrated circuit industry have put a premium on throughput performance. Current lithography tools expose approximately 100 wafers per hour. Each 300-mm-diameter wafer contains about 125 die with an area of 22 × 36 mm2. This c ­ orresponds to a sustained information data rate onto the wafer of ~1 THz! Nonetheless, the cost of optical lithography remains an issue for nanotechnology, both at the research and the early-stage commercialization phases. The impressive and very capable lithography tools used by the integrated circuit industry are very expensive both in initial costs (approximately $25 million) and in operating costs (the typical mask set costs for a modern microprocessor approach $2 million). Clearly these numbers only make sense in a high-volume, high-product-value manufacturing context and are out of reach for a typical research venue or a fledgling product-development endeavor. Fortunately, many applications in nanophotonics require only a periodic pattern and a much sim- pler laboratory-scale technology, interferometric lithography (IL); based on the interference of a small

98 Nanophotonics number of coherent laser beams, IL can produce useful patterns over large areas and large volumes with considerable, but not total, pattern flexibility, and with dimensions that today are approaching the 20 nm scale. IL creates photonic crystal structures in two and three dimensions by using light to f ­ abricate structures that interact with light. The IL pattern can be written into both positive and negative p ­ hotoresists. After exposure, the resist is developed to remove, for example, in the case of a negative resist, the unradiated regions, leaving a bicontinuous air/polymer structure that is rationally designed by choice of the set of beam parameters. Combining multiple interferometric exposures and mix-and-match with lower-resolution, laboratory-scale optical lithography and with a limited use of higher-resolution e-beam lithography dramatically expands the available range of patterns (Brueck, 2005). There are at least two approaches to three-dimensional structures. Structures can be built up in a layer-by-layer fashion very analogous to traditional semiconductor manufacturing. In particular, quite a bit of attention has been devoted to a woodpile structure, with alternating layers of long bars oriented in the x and y directions, with each sequential layer in the same direction offset by one-half the pitch (Ho et al., 1994). Most often the fabrication has used standard semiconductor processing technologies (Lin et al., 1998; Yamamoto, 2006). Alternatively, using multiple-beam, noncoplanar interferometric lithography, the entire three-dimensional photonic crystal pattern can be produced in a single series of exposures in a thick, photosensitive material (Campbell et al., 2000; Shoji et al., 2003; Wang et al., 2003). Many crystal symmetries are accessible by adjusting the number, intensity, and polarizations of the various beams (Cai et al., 2002; Ullal et al., 2003). This interferometric technique has been extended to the fabrication of compound (interpenetrating) lattices such as the woodpile structure and the diamond lattice (Lin and Fleming, 1999; Zhong et al., 2005). However, because of the low-dielectric constant of photoresists, the as-made structure is ineffective as a photonic crystal. One approach is to use the struc- ture as a template for infiltration and then etching/burning out of the polymer to yield a high-dielectric inverse structure (Blanco et al., 2004). Another promising route to very large area photonic crystals is to use phase masks. Here, the phase mask is placed above the photoresist and a single-incident-plane wave source creates a set of exit beams emanating from the phase mask that interfere inside the resist and create a periodic structure. The challenge is to learn how to design the structure of the phase mask in order to create the desired set of beams for the targeted photonic crystal. Phase masks are usually made of polydimethylsiloxane (PDMS) from an initial master (up to 4 inches in diameter, which itself is made by e-beam lithography or by interference holography). PDMS is inexpensive, transparent, and conformal material (self-aligning to the photoresist) that can be used multiple times. Figure 3-8 shows a two-dimensional quasi-crystalline master made by eight successive exposures with 45 degree rotations between exposures, using a Lloyd’s mirror IL set up. The PDMS phase mask is made by replicating the laser, and the three-dimensional quasi-crystalline structure is created by IL from the set of beams launched from the two-dimensional quasi-crystalline PDMS phase mask. The incorporation of design defects into photonic crystals is essential for devices. Laser writing (etching) and focused ion-beam etching and/or material deposition can be used to place defects in two- dimensional photonic crystals. For three-dimensional writing, the use of two-photon polymerization using a confocal microscope with an x,y,z stage drive can be used to serially write defects (Lange et al., 2006). Two-photon approaches that rely on the nonlinear response of materials are an alternative approach to three-dimensional fabrication that have met with considerable success. Photopolymerization is an example that has been used extensively (Cumpston et al., 1999; Sun et al., 1999). The throughput of two-photon approaches can be substantially improved with the use of multiple spots formed by microlens arrays (Kato et al., 2005).

ENABLING TECHNOLOGIES 99 FIGURE 3-8  A novel method for producing large-area three-dimensional nanostructured quasi-crystalline ­materials uses two-dimensional multiple-exposure lithography to produce an octagonal quasi-periodic surface-relief template 3-8 (background image). A replica in polydimethylsiloxane is then used as a phase mask to create three-­dimensional bicontinuous axial quasi-crystalline SU-8 epoxy nanostructures magnified to show subsurface structural details (see insert at left). At bottom right is a corresponding diffraction pattern produced by a narrow laser beam, while at top right is a simulation of a binarized surface-relief template. SOURCE: Bita et al. (2007). Copyright Wiley‑VCH Verlag GmbH & Co. KGaA. Reproduced with permission. Nanoimprinting Nanoimprint technology, replicating a pattern from a three-dimensional mask into a polymer film on a substrate, is a rapidly developing approach that has demonstrated resolutions to about 10 nm. This approach is a very appealing alternative to optical lithography for planar manufacturing (Chou et al., 1997). Issues with this technology involve defining the three-dimensional mask (which is most often made using e-beam lithographic approaches) and the chemical interaction between the mask and the pattern that allows a clean separation and continued reuse of the mask. Nanoimprint lithography has evolved into an important research and commercial-product area with many active participants and a wide range of alternative techniques being explored. A recent review is provided by Guo (2004). Stacking Membrane Structures Gas-phase etch processes such as reactive ion etching (RIE) or inductively coupled plasma etching generally have the commensurate resolution for transfer of the lithographic pattern into the substrate.

100 Nanophotonics a b FIGURE 3-9  (a) Schematic of fabrication process and (b) electron micrographs of fabricated three-dimensional photonic crystal lattice. SOURCE: Reprinted with permission Lin and Fleming (1999). Copyright 1999 IEEE. Early realization of two-dimensional photonic crystal structures in GaAs- and silicon-based materials were achieved through e-beam lithography (Wendt 3-9 1993) or x-ray lithography (Foresi et al., 1997) et al., and RIE processes. Similar processes serve to define the highest-resolution photonic crystal structures currently fabricated, but there are still difficulties in forming truly three-dimensional structures in this way: controlled etching of very high aspect ratio structures remains a challenge, and the various three- dimensional angles and symmetries are not always easily obtained through planar processes. One approach is to form “membrane” structures, where optical confinement in the vertical direction is achieved through the contrast in the index of refraction between the semiconductor (or high index ­material) and air (Reese et al., 2001). Lin and colleagues developed a means of forming three-dimensional photonic crystals by planar processing using a repetitive succession of formation of silicon (Si) nanorods, filling in with silicon dioxide (SiO2), and planarization. Using that kind of processing approach, three-dimensional photonic structures have more recently been produced in tungsten (Li et al., 2003). Another method for fabricating photonic crystals is that of using direct laser writing. There are cur- rently international research efforts, notably in Germany and Japan, using direct laser writing to fabricate three-dimensional nanophotonic structures and other nanophotonic structures. Other top-down processes have been developed, as described in a recent review paper by Lopez (2003). Figure 3-9 illustrates photonic crystal fabrication. Photonic Crystal Fibers Photonic crystal fibers are typically fabricated using a “stack-and-draw” technique (see Figure 3-10) by which silica capillary tubes are assembled into a two-dimensional periodic preform bundle. Photonic

ENABLING TECHNOLOGIES 101 FIGURE 3-10  A stack of glass tubes and rods (a) is constructed as a m ­ acroscopic “preform” with the required photonic crystal structure. It is then fused together and drawn down to fiber (c) in two stages using a standard fiber drawing tower. To soften the silica glass, the furnace (b) runs at 1800°C to 2000°C. SOURCE: Russell (2003). Reprinted with permission from AAAS. 3-10 crystal fiber preforms are fabricated by building a close-packed arrangement of silica tubes around a central silica rod or tube and together these replicate the desired fiber structure. Precision-machined jigs are used to facilitate this process, and the completed preform is typically held together with platinum wire during drawing. The entire stack can then be drawn in a conventional fiber drawing tower to achieve the much smaller, desired dimensions (Russell, 2003). Exceptional precision is essential during the assembly of the preform, together with fine control of all drawing conditions, to ensure that viscous forces do not distort the fiber during formation and to preserve the integrity and geometry of the structure. While modified chemical vapor deposition is not used directly to create the preform, the process can still be used to fabricate the high-purity, fused-silica components that make up the preform. One of the major disadvantages of this stack-and-draw technique is the contamination of glass ele- ments, as only small amounts of dust on the glass surface can result in a significant increase in fiber attenuation as well as leading to fiber breaks during fabrication and subsequent rewinding. Therefore, this fabrication must occur under strict clean-room conditions. In the past, other techniques such as extrusion (Hori et al., 2003) and drilling have been used to fabricate photonic crystal fiber preforms using soft materials such as polymers and soft glass, since with soft materials the structure can collapse during the drawing process. Such techniques are less suitable for producing long fiber lengths. Recently, the stack-and-draw technique has been successfully used to fabricate photonic crystal fibers from sulfur hexafluoride glass (Wolchover et al., 2007), a soft glass with high optical nonlinearity. For the fabrication of polymer-based photonic crystal fibers, a two-stage draw method is typically used (van Eijkelenborg et al., 2003). The primary preforms (typically 50 mm to 100 mm in diameter) are fabricated by drilling the required hole pattern in the polymer material using

102 Nanophotonics computer-controlled milling. The primary preforms are drawn to 5 mm to 10 mm diameters. They are then sleeved 10 about twice this diameter and drawn down to their final size, typically 100 microns to 400 microns in diameter. The first successful photonic crystal fibers were demonstrated using silica in 1995 (Knight et al., 1996). Currently, four companies provide commercial fibers. Directed self-assembly and directed epitaxial growth An important emerging trend is the combination of top-down lithographic approaches with bottom- up self-assembly. Lithography allows hierarchical structures over many length scales, from macro- to nanoscales, but it is having increasing difficulties at the lower end of the nanoscale range (as discussed above). Self-assembly, in contrast, shows its greatest strength at smaller scales (and in short-range correlations) and tends to have more difficulty with long-range order (over many lattice constants). A strategy is to marry the two approaches: define a pattern with top-down lithography and build in the hierarchical length scales, and continue the fabrication with self-assembly processes that integrate with the lithographic structure. This concept is applicable both to traditional self-assembly such as colloidal crystallization and to epitaxial growth (Cheng et al., 2006b). Figure 3-11 shows an example of top-down and bottom-up directed assembly. Polymerization-Induced Phase Separation Polymerization-induced phase separation (PIPS) is an IL technique whereby a miscible mixture of a photopolymerizable monomer and a low-molar-mass liquid crystal are irradiated by multiple coher- ent light beams, and the resultant polymerization causes the liquid crystal to phase separate around the regions of highly cross-linked polymer. The final composite material has an index variation on the scale of the incident wavelength, and depending on the choice of beam parameters it can exhibit one-, two-, or three-dimensional periodicities. Because the liquid-crystal component can be reoriented by an external field, by index matching the polymer matrix with one of the principal indices of the liquid crystal, PIPS samples can be switched to have a stop band or to be transparent to incident radiation. The ability to turn on the index difference also allows beam steering using PIPS materials and is a means to enable sensor protection (Urbas et al., 2004). Nanoscale Crystal Growth (Nanowires) As shown in Figure 3-12, one example of the combination of lithography and epitaxial growth is the recent growth of gallium nitride nanowires in defined positions and without the need for catalysts as is commonly used in the vapor-liquid-solid growth process. This example also provides a case in point of the blurring of the distinctions between fabrication and growth. In this case the starting ­material includes a silicon carbide (SiC) substrate with an epitaxial (MOCVD) GaN buffer layer and a thin (~30 nm) silicon nitride (SiN) selective-growth mask layer. Interferometric lithography is used to pattern a hexagonal array of circular holes in the SiN mask layer (~200 nm diameter and 500 nm pitch). MOCVD growth is then Manufacturers of photonic crystal fibers include Crystal Fiber (Denmark) (see http://www.crystal-fibre.com/); IVG Fiber (Canada) (see http://www.ivgfiber.com/custom.htm); Paradigm Optics (United States) (see http://www.paradigmoptics.com/ structures/fiberstructures.html#nano); and TEGS Ltd. (Russia) (see http://www.tegs.ru/en/apps/pcf.shtml). Last accessed on April 15, 2007.

ENABLING TECHNOLOGIES 103 FIGURE 3-11  Example of top-down and bottom-up directed assembly. Templating of spherical block polymer domains within one-dimensional templates of varying width and scanning electron microscope micrographs of ordered arrays of spherical domains with N = 2 to 12 rows. SOURCE: Adapted with permission from Cheng et al. (2004). Copyright 2004 by Macmillan Publishers Ltd. Top View Side View FIGURE 3-12  Scanning-electron-patterned gallium nitride (GaN) nanowires. The hexagonal pattern of the n ­ anowires is created by using a mask layer on top of single-crystal GaN. SOURCE: Reprinted with permission from Hersee et al. (2006). Copyright 2006 American Chemical Society. 3-12

104 Nanophotonics used to grow the nanowires directly; in contrast to the nanowires described above, no catalyst is required (and no unintentional doping occurs), and the nanowires are precisely defined in size and position. Findings Finding 3-1. Developments in synthesis, growth, and fabrication for photonic nanostructures extend across a wide range of materials and techniques and follow nontraditional paths. While it is tempting to assume that those countries which today have an extensive infrastructure for traditional photonics and microelectronics will continue to be the dominant developers of this new technology, this infu- sion of new ideas and new technologies means that new players can emerge over the 10-to-15-year time frame covered by this report. Finding 3-2. Traditional electronic-device fabrication employs well-defined and largely separate stages of synthesis, growth, and fabrication. In contrast, the generation of nanophotonic materials and devices blurs these distinctions, and certainly the order of these stages, interleaving them in new and novel ways. Finding 3-3. Nanophotonic devices increasingly incorporate a wide range of materials and pro- cessing methods. Photonic crystals made of or incorporating organic materials are likely to provide sensitivity and specificity for the detection of chemical and biological agents. Modeling and Simulation in Nanophotonics One major reason for the rapid progress in the field of nanophotonics globally is the increasing avail- ability of powerful computational methods for the design and simulation of nanophotonic structures, devices, and systems. Because of the pervasiveness of electromagnetic phenomena and the necessity of understanding radiation-matter interactions, a broad set of computational tools that have applicability to nanophotonics has been developed. An interesting feature of electromagnetic interactions with dielectric material is the scalability of the solutions of Maxwell’s equations. For example, if one solves for the characteristic modes (wave functions) of a photonic crystal with characteristic lattice constant a, and frequency ω, and if one scaled up that structure by a factor s, to form a new lattice constant sa, then the characteristic resonant frequencies would similarly scale up to values of sω. Although first developed for radio-frequency modeling and analysis, computational approaches for solving Maxwell’s equations scale directly to optical frequencies and subwavelength structures in this regime, with appropriate modifica- tions in materials parameters. Scalability was critically important in the earliest experimental exploration of photonic crystal structures, with critical dimensions of millimeters, probed at microwave frequen- cies. This phenomenon established the validity of predictions made for current-day 100 nm structures (Yablonovitch and Gmitter, 1989). Since most nanophotonic elements can be modeled using Maxwell’s equations to a sufficiently high degree of quantifiable accuracy, the existence of predictive design and simulation capabilities in this field is uniquely mature relative to other nanotechnology disciplines. This section discusses some of the most popular computational methods currently in use in nanophotonics.

ENABLING TECHNOLOGIES 105 Finite Element Method Historically, scalar finite element methods (FEMs) were initially developed by and for civil and mechanical engineers for the analysis of structural and materials problems. However, since FEM is useful for the solution of sets of partial differential equations, its development as a numerical modeling method for electromagnetics has been very successful. With the availability of powerful computers and commercial finite element software packages, three-dimensional electromagnetic modeling of complex systems and structures by this method is now a major tool in the development of nanophotonics. In finite element analysis, the computational space is divided into a number of small, homogeneous pieces, or elements. The model contains information about device geometry, material constants, incident field properties, and boundary constraints. The element size can be varied; it can be small where geometric details exist and much larger elsewhere. In each finite element, a simple, frequently linear, variation of each field quantity is assumed. The goal of any finite element analysis is to determine the field quantities at the nodes, literally the corners of the individual elements. Generally, finite element analysis techniques solve for the unknown field quantities by minimiz- ing an energy functional, describing all the energy associated with the configuration being analyzed. In order to obtain a unique solution, it is necessary to constrain the values of the field at all boundary nodes. A major weakness of the finite element method is that it is relatively difficult to model open con- figurations (i.e., configurations in which the fields are not known at every point on a closed boundary). V ­ arious techniques such as absorbing boundaries are used in practice to overcome this characteristic. An advantage that FEMs have over other electromagnetic modeling techniques stems from the fact that the electrical and geometric properties of each element are defined independently. This permits the problem to be structured with a large number of small elements in regions of complex geometry and fewer, larger elements in relatively open regions. It is therefore possible to model configurations with complicated geometries and many arbitrarily shaped regions in a relatively efficient manner. One major reason for the popularity of FEM for electromagnetic problem solving is the ­availability of a highly popular, commercially available software package, COMSOL Multiphysics (formerly Femlab). COMSOL is based on FEM code developed within the research group of the late Germund Dahlquist of the Royal Institute of Technology in Stockholm, Sweden. This tool, rapidly becoming a world standard in both academia and industry, provides FEM solvers for electromagnetics as well as all major engineering disciplines, including heat transport and fluid flow, in a user-friendly environment that allows for the simultaneous solution of multiple coupled problems, such as electromagnetics and heat dissipation in a device structure. COMSOL is currently used by scores of nanophotonics research groups worldwide. Finite-Difference Time-Domain Method A particularly widely used modeling technique is the finite-difference time-domain (FDTD) method, initially described by Yee and given its acronym by Taflove (Taflove, 1980; Yee, 1966). Since Maxwell’s equations relate time-dependent changes in the electric (magnetic) field to spatial changes in the magnetic (electric) field, FDTD programs implement an explicit time marching algorithm for solving Maxwell’s curl equations on a spatial grid (Sullivan et al., 2000). Modeling a new system is reduced to grid genera- tion, instead of deriving geometry-specific equations. Thus, previously uninvestigated and potentially complex geometries do not require new formalisms. The time marching aspect of the method allows one to make direct observations of both near- and far-field values of the electromagnetic fields, and at any time during the simulation. With these “snapshots” the time evolution of the electromagnetic fields can be calculated and visualized (Lazzi and Gandhi, 2000; Young-Seek et al., 2000).

106 Nanophotonics Because FDTD is a time-domain technique, when a broadband pulse (e.g., a Gaussian) is used as a source, a single simulation can model the response of the system over a broad range of frequencies. This is highly useful for simulations for which the resonant frequencies are not known precisely (such as in the modes of a photonic crystal waveguide or resonator). A typical implementation of this method, for example, may involve an incoming light pulse or wave that illuminates a target containing one or several nanoparticles or structures. The electromagnetic fields excite plasmons or polaritons in the individual nanoparticles. The plasmons on the individual nanoparticles interact, resulting in a complicated and time-dependent electromagnetic field. Since the dimensions of the composite target may be of the same order of magnitude as the wavelength of the light, retardation effects are included naturally. Thus, the FDTD method is well suited to transient analysis problems and to modeling complex inhomogeneous complex nanostructures of arbitrary geometry, as well as arrangements of multiple nanostructures (Yu et al., 1997). With the implementation of periodic boundary conditions, for example, this method is extremely useful for modeling photonic band-gap-based structures and devices. The FDTD method has several features that make it a highly desirable algorithm. It is fully retarded and fully explicit, which reduces the computational overhead needed to solve a particular problem. The sources of errors in the method are well known and bounded (Kashiwa et al., 2003; Wang and Teixiera, 2004). � The dielectric function, ε (ω , r ), is used to simulate the materials in the grid. A value for e is specified at every grid point. For a realistic modeling of the optical properties of nanostructures, it is important to use experimentally measured dielectric functions appropriate for the nanostructure. The use of the electric flux density form of Maxwell’s equations provides two major benefits: a simpler time-domain implementation of dielectric functions and programming modularity. The electric and magnetic field update equations are independent of the choice of dielectric function. Thus, if an object to be modeled requires a different dielectric function, only the electric field update equations need to be modified. Additionally, the method is easily parallelizable owing to the local nature of finite differences. This localized aspect enables the use of simple parallel methods, on either shared or distributed memory systems. For example, the FDTD method can be parallelized using domain decomposition, where the computational grid can be broken into several smaller grids, where each subgrid is updated by a different central processing unit (CPU). To complete the update, subgrids only need to exchange boundary values with adjacent subgrids. This type of functionality enables faster and more efficient use of computational resources. The possibility of using variable grid size drastically reduces the data size and execution time and, more importantly, enables the modeling of larger and more complex nanostructures and nano­ structure aggregates (Amon et al., 2006). Boundary Element Method The boundary element method (BEM) is a method of moment approach applied to the solution of surface integral equations. BEM is outstanding in analyzing unbounded radiation problems, perfect conductors, extended configurations, and homogeneous dielectrics. It is best suited for modeling thin or long wires or surfaces, for substrates, boundaries, or layered media, and as such BEM complements the FDTD code previously described. The method solves the integral form of Maxwell’s equations on the discretized boundary of the intended problem (Massoud and White, 2002a; Massoud and White, 2002b; Matsuhara et al., 1991; Taotao et al., 2004). The surface discretization in BEM leads to fewer numbers of unknowns than those in volumetric methods. Volumetric methods usually generate a global mesh for all parts of an analyzed structure and for surrounding external space. This causes the number of unknowns to increase significantly. Solving this large generated linear system requires excessive memory

ENABLING TECHNOLOGIES 107 and consumes much CPU time, which makes the analysis of complex three-dimensional nanostructures sometimes impractical. Gaussian elimination is a standard method to solve linear systems resulting from BEM formulation, but it is computationally expensive since it requires on the order of n3 operations (Bängtsson and Neytcheva, 2005). Other Numerical Methods Various other numerical methods are currently used in the analysis of electromagnetic fields in microstructured materials with complex geometry. They include the T-matrix method and its variants, which are multiple scattering approaches, formulated in terms of interior and exterior spherical harmonic expansions; and the discrete dipole approximation and its variants, essentially volume integral equa- tion methods discretized using a simple quadrature rule on a regular lattice (Draine and Flatau, 1994; Mishchenko et al., 2002). An emerging new approach is fast multipole methods (FMMs) based on boundary integral equa- tions. Boundary integral equations have the advantage that, by placing the degrees of freedom on the dielectric (or metal) interface itself, it is possible to construct fast, robust, high-order accurate solvers. One disadvantage is that FMMs require significantly more complex numerical algorithms. From a historical perspective, despite their advantages in terms of accuracy and robustness, integral equations were often avoided in favor of other methods because they led to dense linear systems for which fast algorithms were not available. In the past decade, progress was made in many areas, and the state of the art now combines high-order accurate surface discretizations, high-order accurate quadratures, and multilevel fast algorithms. In an iterative solution procedure, the fast multipole method, for example, reduces the cost of applying the integral operator from O(n2) to O(n log n), n is the number of nodes in the discretization of the domain boundaries (Cheng et al., 2006a; Chew et al., 2001; Contopanagos et al., 2002; Darve, 2000; Greengard et al., 1998). These new methods have yielded speedups of several orders of magnitude, bringing previously intractable problems within reach of modest computing platforms. Electromagnetic scattering from objects with millions of points in the surface discretization can be computed in minutes or hours. FMM-based schemes now lie at the heart of some of the leading chip-design software packages and have dramatically changed the standards for capacitance extraction and other “physical verification” tools. Their incorporation into metamaterial optics is likely to have an equally important effect—that of enabling accurate simulation of large-scale, physically realistic systems. Analytic Methods While not as generally applicable as the techniques described above, various analytic approaches to solving Maxwell’s equations for various geometries and materials continue to be widely used because of the physical insight that they provide and because of the ability to vary the parameters of the design over wide ranges with only minimal computing time, again for purposes of physical insight. The most widely used of these analytic approaches is the rigorous coupled wave analysis (RCWA) first introduced by Gaylord and Moharam (1985) for grating diffraction problems. The basic idea of RCWA is to expand the fields in the regions away from the grating and in the grating region in a F ­ loquet expansion (infinite series with the wave vector in the direction of the grating varying by integer m ­ ultiples of the grating period). The electromagnetic properties of the grating (the permittivity in the case of a simple metal grating) are similarly expanded and substituted back into Maxwell’s equations. This procedure results in coupling between the various terms in the Floquet expansion. The series is

108 Nanophotonics then truncated with appropriate consideration for convergence, and the result is a simple linear equation eigenvalue problem for which there are many available solution approaches. This approach has been extended to two-dimensional gratings and to negative index materials and plasmonic aperture array prob- lems (Minhas et al., 2002). Pendry has introduced the transfer matrix approach, a related calculational procedure (Pendry and Bell, 1996). Another set of tools arises from the bandstructure calculations of solid-state physics. One specific example is the Kronig-Penney model for a one-dimensional grating. In this case, Maxwell’s equations are solved exactly in the grating region by matching boundary conditions across the teeth of the grating, and these exact solutions are again expanded in a Fourier series and matched to the Floquet expansions above and below the grating. This again leads to a truncated linear algebra eigenvalue solution. The advantage of this approach is that it directly provides the modal solutions in the grating region that can provide additional physical insight. In the field of plasmonics, plasmon hybridization is one important analytical approach that has been highly useful in determining the resonances of complex nanoparticles and nanostructures, including the coupling between localized plasmons of nanoparticles and the propagating plasmons on macroscope films and wires (Prodan et al., 2003; Wang, 1991). It has recently become apparent that the plasmons of metallic nanostructures, while describable by classical electromagnetic theory, exhibit certain char- acteristics that are analogous to electrons in quantum systems. This is seen most clearly in complex nanostructures, where plasmons on neighboring structures or surfaces interact, for then the plasmons mix and hybridize just like the electron wave functions of simple atomic and molecular orbitals. This property, termed plasmon hybridization, governs the optical properties of metallic nanostructures of increasingly complex geometries, providing the scientist with a powerful and general design principle that can be applied to guide the design of metallic nanostructures and to predict their resonant properties. Although similarities between plasmons in nanostructures and atomic and molecular wave functions have long been a casual observation of workers in this field, it is only recently, where breakthroughs in the controlled chemical fabrication of metallic nanostructures of various shapes and sizes have been combined with powerful computational methods, that this analogy has been realized and verified, and subsequently exploited in the design of complex plasmonic nanostructures. Plasmon hybridization theory deconstructs a composite nanostructure into more elementary shapes and then calculates how the ­plasmon resonances of the elementary geometries interact with each other to generate the hybridized plasmon modes of the composite nanostructure. This theory enables scientists to draw on decades of intuition from molecular orbital theory to predict the plasmonic response of complex nanostructures correctly. Characterization Techniques for Nanophotonics Developing new techniques for the characterization of the optoelectronic properties of materials on a nanometer scale is essential for nanophotonics. There are various methods now under development for such characterization capability. Advanced Microscopies The wavelength of light limits the ultimate size of an object that we can see directly using con- ventional optics. Thus, even with the best optical microscopes, it is not possible to resolve objects less than approximately 400 nm apart, because the wavelength of the light is longer than the separation, and the two objects therefore appear blurred together. One route to improved resolution is to reduce the wavelength of the light used for viewing. Shorter-wavelength light, corresponding to the UV and x-ray

ENABLING TECHNOLOGIES 109 regions of the electromagnetic spectrum, provides better resolution than that possible with conventional optical microscopes. To date, soft x-ray microscopes have achieved a resolution of a few nanometers. In principle, shorter-wavelength, harder x-rays should allow even better resolution, but to date such micro- scopes have been limited to a resolution of about 20 nm. This resolution is limited not by the wavelength but rather by the ability to fabricate the appropriate x-ray lens (known as a zone plate) that is used to focus the x-rays (Chao et al., 2005). Advances in nanofabrication techniques will one day override this limitation, allowing x-ray microscopes with subnanometer resolution to be built. Other characterization techniques for nanophotonics based on linear and nonlinear optical spectroscopies, both in the time and the frequency domains, are described in Chapter 2 of this report, in the section entitled “Techniques for Imaging and Spectroscopy of Plasmonic Structures.” Other nonlinear optical microscopies can be used to provide even more information. For example, nonlinear optical techniques such as multiphoton excitation and Raman spectroscopy are now being used to image specific chemicals or nanostructures, with three-dimensional submicron resolution, and to study time-dependent processes involving these species. Such techniques are useful for the characterization of three-dimensional nanophotonic structures. Scanning Probe Microscopy Scanning probe microscopy (SPM) covers several related technologies for imaging and measuring surfaces on a nanometer scale. SPM technologies share the concept of scanning an extremely sharp tip (3 nm to 50 nm radius of curvature) across the object surface. The tip is mounted on a flexible cantilever, allowing the tip to follow the surface profile (see Figure 3-13). See, for example, Sakurai and Watanabe (2000). FIGURE 3-13  The movement in scanning probe microscopy to map out the surface profile of an object. SOURCE: Committee member Antoinette Taylor, Los Alamos National Laboratory, 2007. 3-13

110 Nanophotonics When the tip moves in proximity to the investigated object, forces of interaction between the tip and the surface influence the movement of the cantilever. These movements are detected by selective sensors. Various interactions can be studied, depending on the mechanics of the probe. Following are the three most-common scanning probe techniques: • Atomic force microscopy (AFM): AFM measures the interaction force between the tip and surface. The tip may be dragged across the surface or may vibrate as it moves. The interaction force will depend on the nature of the sample, the probe tip, and the distance between them. As an example, consider having a probe only a nanometer in diameter scanning across some molecules sitting on a surface. The displacement of this probe as it crosses the molecules, similar to that of the tip of an atomic force microscope, allows one to create a nanoscale map of the surface roughness by measuring the deflection of an ultrafine tip on a cantilever as it is “dragged” across the surface. The atomic force microscope is but one of a wide range of scanning probe microscopies, each of which relies on a different local nanoscale effect for its contrast mechanism. • The scanning tunneling microscope (STM): The STM is a particularly powerful tool that relies on the measurement of tiny electrical currents that pass between the tip and the surface, with changes in these currents reflecting the local densities of electrons on the surface. Such STMs have been used to map out the positions of individual atoms on the surface of a crystal, as well as the posi- tions of atoms within a molecule. Perhaps even more exciting, the tips of these STMs have been used to move and position individual atoms sitting on a crystal surface, allowing the creation of controlled structures with atomic dimensions. This approach has also been pursued using ultra- fast lasers, allowing the study of time-dependent optical phenomena with unprecedented spatial resolution (Yarotski and Taylor, 2004). • The near-field scanning optical microscope (NSOM): The NSOM scans a very small light source very close to the sample. Detection of this light energy forms the image. The NSOM provides subwavelength resolution that is well below the conventional limit of optical microscopy. In one form of NSOM, the tip used in an atomic force microscope or an STM is replaced by an optical fiber tapered to a tip approximately 50 nm across; this tip is positioned a few nanometers above the surface of interest and then scanned across it. Laser light is sent down the fiber and tunnels out of the tip, illuminating the sample over an area approximately equal to the diameter of the tip. Scattered light or fluorescence from the sample is then detected, allowing an optical image with approximately 50 nm resolution when a laser wavelength of 500 nm is used. The image in Figure 3-14 shows an NSOM image employing femtosecond white light sources to perform extinction measurements of individual gold colloids and assemblies of gold nanoparticles. These SPM approaches have also been pursued using ultrafast lasers (both an ultrafast STM and an ultrafast NSOM), allowing the study of time-dependent optical phenomena with unprecedented spatial resolution. Many additional new techniques, such as multiphoton microscopy and ultrafast confocal microscopy, are now being developed to apply optical methods to probe nanoscale and even single- molecule processes in exquisite detail. Scanning Electron Microscopy The scanning electron microscope (SEM) is capable of producing high-resolution images of the sample surface. Acquired images have a three-dimensional appearance and are useful in determining the surface structures of the sample. A focused beam of electrons scans the surface producing secondary

ENABLING TECHNOLOGIES 111 143.30 nm 5 µm 5 µm 2.5 µm 2.5 µm 0.00 nm 0 µm 0 µm 5 µm 0 µm 2.5 µm 5 µm 0 µm 2.5 µm FIGURE 3-14  Topographic image (left) and near-field scanning optical microscope (NSOM) image (right) of assemblies of gold nanoparticles. SOURCE: Victor Klimov, Los Alamos National Laboratory, 2006. 3-14 electrons that are registered with the detector in a line-by-line fashion. Electronics then captures all lines, displaying them on the screen as an image. When the surface is struck with electrons, other effects also take place. The nature of the SEM’s probe, energetic electrons, makes it uniquely suited to examining the optical and electronic properties of materials. Typical spatial resolution of a high-end SEM is on the order of 2 nm to 5 nm. Figure 3-15 shows a typical SEM and images of three- and two-dimensional photonics lattices fabricated using standard processing techniques. FIGURE 3-15  A scanning electron microscope (left) and a typical image of the three-dimensional photonic lat- tice (middle) and the two-dimensional photonic lattice (right). The scale (bar) in the middle panel is 5 microns. The scale (bar) in the right-hand panel is 100 nm. SOURCES: Left: JEOL Company, reprinted with permission; middle, right: Elshan Akhadov, Los Alamos National Laboratory, 2006. 3-15

112 Nanophotonics Transmission Electron Microscopy In the transmission electron microscope (TEM), a beam of electrons is transmitted through a specimen and detected by a sensor yielding information about the material’s internal structure. Unlike scanning electron microscopy, TEM samples need to be prepared (thinned down) in order to be useful for measure- ments. With the typical resolution of 0.1 nm, TEMs have found great use across many fields of science: biological samples and materials could be investigated at much finer length scales, allowing for investi- gation of hidden internal lattice defects and shedding light into areas inaccessible with other techniques. Figure 3-16 illustrates the typical TEM instrument and two images acquired from quantum dots. Nanophotonics Devices Wavelength-Scale Devices Wavelength-scale devices include photonic crystal-type devices that have a periodicity on the order of a vacuum wavelength, but they do incorporate precision at the nanometer scale, justifying the nano- photonic appellation. Of particular note is the success of fabrication methods for photonic crystal fibers. A stack-and-draw technique assembles silica capillary structures into bundles, forming photonic crystal structures; the capillaries are fused together; and then the entire structure is drawn out to form a much-smaller-diameter fiber with the correct structure (Lim et al., 2004). The first successful photonic crystal fibers were formed in 1995, and the committee believes that the ease in fabrication has led to the rapid commercialization and availability of photonic crystal fibers. Conventional semiconductor processing has emerged as a practical method for the creation of nano- photonic structures. The ability to control light down to a level smaller than a quarter-wavelength is the hallmark of nanophotonic structures. The canonical wavelength for optical communications and signal processing is λ = 1,550 nm. But allowing for the typical refractive index of semiconductors, n ~3.5, the actual wavelength inside a material is only ~450 nm. Reducing this to a quarter-wavelength results in a critical dimension for FIGURE 3-16  A transmission electron microscope (left) and images of a quantum dot used in photonics applica- tions (each dot in a periodic array represents an atom). SOURCE: left: Image courtesy of FEI Company; center and right: Paul Alivisatos, University of California, Berkeley, 2007. 3-16 a,b,c

ENABLING TECHNOLOGIES 113 ITRS Roadmap: 100nm FIGURE 3-17  Minimum refractive size on semiconductor devices decreases steadily, per Moore’s law. NOTE: ITRS, International Technology Roadmap for Semiconductors. SOURCE: Van den hove (2002). Reproduced with permission. 3-17 original figure replacing new figure nanophotonics of about 100 nm. This has been a challenging size scale to control, but the semiconductor 9/19/07 industry passed that dimensional threshold in 2003. See Figure 3-17. Moore’s law tracks the minimum feature size that the semiconductor industry has been able to manufacture commercially. Since 2003, all of the usual optoelectronic components and devices needed for optical routing of signals have been demonstrated in a commercial silicon foundry manufacturing process. All, that is, except for the light source itself, which remains external to the silicon but coupled to it by means of lithographically produced nano-optical grating couplers. These lithographic grating couplers have now demonstrated mode-matching and insertion loss of less than 1 decibel (dB) and are part of the normal process flow in integrated circuit manufacturing. Indeed these nanophotonic chips are also fully electronic chips, since they are made in the same process flow that produces normal integrated circuits. Thus, full optoelectronic integration is achieved, and the nano- optics is accommodated without any changes in the normal operation of a silicon fabrication plant. A nanophotonic chip is a silicon chip that is made in a modern silicon foundry, taking advantage of the fact that a modern silicon foundry can control the position of the silicon/silicon dioxide boundary to within less than 5 nm, justifying the name “nanophotonic.”

114 Nanophotonics Among the individual components that have been demonstrated are optical modulators that run at 20 gigabits per second (Gbps) integrated with driver electronics that currently run at 10 Gbps. Wave- guides, splitters, frequency filters, and high‑quality (Q) resonators have all been demonstrated and are now reduced to a library of devices, whose graphics files can be called up by an optoelectronics system designer (Kalra and Sinha, 2005; Sohler et al., 2005). Selective-area grown epitaxial germanium on silicon has become a standard part of silicon process- ing in modern foundries, largely exploiting the strains that are induced, which increase the mobility and speed of transistors. Serendipitously, the germanium now permits high-speed photodetectors, as part of the standard silicon process. Moreover, these integrated photodetectors are probably the first nanophotonic integrated component that is actually superior to the best discrete component. The reason is that the electronic preamplifier can be part of the photodetector, diminishing detector capacitance. This saves many decibels in optical communications link margin, permitting performance competitive with the best discrete optical communications systems. Thus, the entire suite of optoelectronic signal processing components has recently become available as a standard commercial silicon electronic manufacturing process, and has become a standard optoelectronic manufacturing process. An example is illustrated in Figure 3-18. Flip-chip bonded lasers Silicon Optical Filters - DWDM Silicon 10G Modulators wavelength 1550nm electrically tunable driven with on-chip circuitry highest quality signal passive alignment integrated w/ control circuitry low loss, low power consumption non-modulated = low cost/reliable enables >100Gb in single mode fiber Complete 10G Receive Path Ge photodetectors trans-impedance amplifiers output driver circuitry The Toolkit is Complete • 10Gb modulators and receivers Fiber cable plugs here • Integration with CMOS electronics • Cost effective, reliable light source • Standard packaging technology Ceramic Package 4 FIGURE 3-18  Luxtera complementary metal oxide semiconductor�������������������������������������� ����������������������������������������������������������������������������� (CMOS) photonics technology. SOURCE:� Gunn (2007). Reproduced with permission of Luxtera, Inc. fig 3.18

ENABLING TECHNOLOGIES 115 Deep Subwavelength-Scale Nano-Optical Devices Deep subwavelength-scale nano-optical devices are enabled by the new-found recognition that metallic wires and waveguides can be used to control and focus light down to the scale length of a few nanometers. Such a device has already existed for a long time, in the form of the near-field scanning optical microscope, a commercial imaging instrument in the form of a nanoscopic optical pinhole. New types of focusing structures that resemble tapered waveguides have emerged, allowing for efficient optical focusing to the nanoscale. These focusing structures are in turn fed by some type of antenna structure that converts a free-space laser beam to optical frequency currents and voltages along the metal surfaces. The optical antenna structures can be in the form of dipole antenna, monopole antenna, slot antennas based on Babinet’s principle, grating couplers that resemble Yagi rooftop televi- sion antennas, and so on. Among the recent insights in this field is the recognition that well-designed focusing structures can focus light down to dimensions of a few- nanometers with reasonable efficiency. Some examples of these optical nano-antennas are shown in Figures 3-19 and 3-20. One of the important applications for these nano-optical focusing structures is for concentrating infrared radiation to improve the signal-to-noise performance of infrared sensors. Another important application is for localized laser heating to accomplish heat-assisted magnetic recording (HAMR), and for other, more speculative applications such as maskless scanning lithography. On the magnetic storage roadmap, HAMR is to appear commercially within about 5 years. Among the many issues in making HAMR practical are the many options remaining for the antenna and focusing structure. It is not clear which of these focusing options, represented in Figures 3-19 through 3-21, will emerge as the best for the HAMR application. Packaging and Integration Although the importance of packaging and advanced integration for future nanophotonic technolo- gies will vary vastly with the specific applications, the Committee on Nanophotonics Accessibility and Applicability believes generally that packaging and integration will play a key role in the enablement of nanophotonics. Because nanophotonics is still an infant technology with much potential but at present relatively few specific applications, and because packaging and integration comprise a very broad subject with many different details depending on the particulars of the respective applications, the committee limits the dis- cussion here to the “first level” of packaging technologies, thereby excluding “system-level” packaging aspects. In addition, this discussion focuses on technologies, which enable the integration of optical and electrical devices (e.g., remote sensing, night vision, computing, communications, eavesdropping, and so on) because in these areas packaging and integration technologies are particularly critical to enabling the respective nanophotonic applications. Specifically, military applications, which could be enabled by the combination of (nano)photonics and high-performance complementary metal oxide semiconductors (CMOSs), may include supercomputers on a chip for massive parallel and high-throughput real-time data processing for purposes such as the control of rockets and unmanned vehicles or for uses such as data mining (Kumagai, 2001). The types of applications often entail solving large sets of linear equa- tions and performing many matrix operations as well as Fourier transform analysis, and they will thus require unmatched computing capabilities in conjunction with enormous data rates—potentially larger A Yagi rooftop television antenna is the standard type of rooftop television antenna that has a periodic array of metallic elements.

116 Nanophotonics hν hν λ 4 λ 2 Bow-tie antenna Optical STM half-wave dipole quarter-wave monopole FIGURE 3-19  Two different options for an optical antenna structure that would capture optical waves and focus them to a small spot for heat-assisted magnetic recording. It is not all clear which of these designs or which other options are best. SOURCE: Eli Yablonovitch, University of California, Los Angeles, 2007. Reproduced with permission. 3-19 Ag Ag E(ω) c E(ω) Ag Ag Complementary C-slot antenna Bow-tie slot antenna FIGURE 3-20  Some slot antenna options that are being promoted for capturing free-space lightwaves and focus­ing them to produce a high optical electric field E(ω). These options have more metal than do the ­antennas in ­ Figure 3‑19, suggesting that they might possibly be more efficient against resistive losses. SOURCE: Eli Y ­ ablonovitch, University of California, Los Angeles, 2007. Reproduced with permission. 3-20

ENABLING TECHNOLOGIES 117 e iss ion lin transm far field from conventional lens red e tap dimple lens Ag ou t-c ou pl in g ed Ag ge in-coupling grating FIGURE 3-21  The dimple lens takes advantage of the wave properties of plasmons as they propagate down to the nanoscale. The net effect is somewhat analogous to a tapered parallel plate transmission line (upper left). Progres- sively higher wave impedance toward the vertex of the dimple ensures high efficiency, relative to the inevitable resistive losses in the metallic components. SOURCE: Eli Yablonovitch, University of California, Los Angeles, 2007. Reproduced with permission. 3-21 than 50 terabits per second, which probably only (nano)optics closely integrated with high-performance CMOS will be able to deliver. Technology Environment Nanophotonics—at least in its current form—relies heavily on semiconductor and CMOS fabri- cation capabilities and their associated package and integration technologies. In fact, nanophotonics depends—at least to some extent—on the continued progression of semiconductor manufacturing and integration capabilities as projected by the International Technology Roadmap for Semiconductors (ITRS) (International Roadmap Committee, 2005). Because of this strong dependence, some relevant semiconductor technology trends are discussed here, as these will be driving new packaging and integra- tion technologies, which could be very important for the general enablement of nanophotonics. Because of increasing technology challenges, discussed in more detail below, the committee consid- ers an increasing likelihood of diminishing returns on investment in the semiconductor industry. Such a trend could result in the possibility of fewer investment dollars in the further development of these manufacturing capabilities, which could have potentially important implications for the general devel-

118 Nanophotonics opment of future nanophotonic devices (especially in the optical frequency range). This is particularly true if nanophotonics fails to identify applications that are commercially important enough to warrant the major investments necessary to improve current semiconductor manufacturing capabilities. The tight interdependence of nanophotonics on semiconductor manufacturing will possibly make nanophotonics much more readily accessible in foreign countries (especially India, China, and Taiwan, for example), as some segments of the semiconductor industry may commoditize, which warrants a continued reduction in cost by offshoring and outsourcing. In the past there were two major incentives for the semiconductor industry to keep shrinking the size of the transistor (“transistor scaling”), creating a “win-win” situation for clients and manufactur- ers. Not only did the performance of the transistor increase significantly with every technology node (about 15 percent per generation), but also the actual cost per transistor decreased (typically by gradually increasing the wafer sizes). In addition, smaller transistors enable higher transistor package densities, resulting in more transistors per die and thus even more performance and functionality on the chip level. While there is still a significant economic benefit to be gained from shrinking transistor sizes further, the performance improvements have been lagging recently and the rate of scaling may be slow- ing down—mostly because of the inability to scale the gate dielectric thickness in a power-constrained environment. Today the semiconductor industry is facing two major technical challenges. First, the combination of higher transistor densities and the emergence of leakage current (due to short-channel effects and other causes) has led to an explosion of power dissipation on CMOS circuitry (Frank, 2002). Today’s semiconductor chips are generally considered power-limited, and no additional power consumption can be tolerated (Horowitz, 2007). The second major technology challenge originates from the emergence of “physics-based” or pure statistical variations (Bernstein et al., 2006). For example, the numbers of ­dopant atoms within a transistor channel approaches levels at which the location of individual dopant atoms starts to matter. For example, for 100 dopant atoms (with a current manufacturing scheme of implanting), the statistical dopant fluctuation from device to device is already about 10 percent ( N / N ), affecting transistor threshold voltages (Stolk et al., 1998). Other “physics-based” variations include line-edge roughness or contact resistances. While major technology innovations such as high-k dielectrics will play a key role in the mitigation of the power crisis and novel device structures will be critical to reducing the variability impact, it is generally believed that the semiconductor industry has to find additional ways of enhancing performance besides traditional device scaling (Chen et al., 2006). This trend opens up significant opportunities for nanophotonics and nanophotonic integration. New materials, novel device structures, innovative designs (such as Fin-field-effect transistors, ultrathin silicon on insulator [SOI]), and new CMOS processes will be the key drivers for improving device performance and thus will be more rapidly and readily incorpo- rated. For example, the semiconductor industry has already started aggressively to adopt more materials and elements into the CMOS process, redefining CMOS compatibility. While before the 1990s all CMOS circuits were made of only seven elements (silicon, hydrogen, arsenic, boron, oxygen, aluminum, and phosphorus), today semiconductor devices have many more elements (more than 15), which could enable and accelerate CMOS-compatible nanophotonic integration, as nanophotonics undoubtedly requires a more diverse selection of elements and materials. The continued challenges with respect to increasing power and variability and the insatiable demand for more computing performance have led to the rise of multicore processors, in which instead of one very high power density and speed processor core with a deep pipeline, several cores at less frequency and power are implemented, yielding more performance per unit of power on the chip level. The ­numbers of cores on a single die chip will be drastically increasing in the near future. For example, the ­International

ENABLING TECHNOLOGIES 119 Business Machines (IBM) cell processor already has nine different processor cores, and the trend is expected to be continued (Kahle et al., 2005). These emerging chip multiprocessors (CMPs) are fundamentally different from past unicore micro- processors because their performance is much more determined by the efficiency and throughput of the global intrachip communications infrastructure. However, conventional electrical interconnects are starting to reach limitations such as increased latency; increasing power consumption (because of the power consumption of multiple repeaters, which worsens the power problem of an already power- l ­ imited microprocessor); and resonant-cavity bandwidth limitations to an extent that only a small fraction of the die can be reached within one clock cycle (time-of-flight problem). As a consequence, the semi­conductor industry has started looking at more densely integrated optical interconnects and nanophotonic solutions, which could enable vastly superior, higher-throughput, distance-independent, potentially less-power-­consuming and less-latency-prone on-chip communication systems. This ­activity is consistent with a general trend in the computing industry, which has been gradually introducing optics in its system getting closer and closer to the microprocessor (see the trend in Figure 3-22). This trend is very important because it may imply that optics will be introduced closer and closer to the chip level, making nanophotonics devices commercially highly relevant. As for the chip design, and in sharp contrast to the past when a high-performance core was individu- ally custom-designed, CMPs are using more modular system-on-chip (SoC) design techniques in order to deal with the increased complexity at growing design costs. It is expected that these more modular design strategies will more readily enable outsourcing and offshoring of circuit design and possibly eventually fabrication work, which could enhance the accessibility of nanophotonics in foreign countries. FIGURE 3-22  Evolvement of the role of optical communication in server computing systems. ���������� NOTE: MAN/ WAN, mobile area network/wide area network; SONET, s������������������������������������������������� ynchronous optical networking�������������������� ; ATM, a������������ synchronous transfer mode�������������������������������������������������������������������������������������������� ; LAN/SAN, local area network/storage area network; PCI, peripheral component interconnect�� ����������������������������������� . 3-22 SOURCE: Benner et al. (2005). Reprodueced with permission of IBM Research.

120 Nanophotonics Two very important implications can be gleaned from these general semiconductor technology trends. First, because of power limitations and statistical variability, the semiconductor industry is undergoing a major paradigm shift to multicore processing with increased on-chip communication requirements. This shift is quite likely to accelerate the introduction of (nano)photonic solutions, which provide an attrac- tive alternative to traditional electronics. Second, the continuously increasing technological challenges associated with exploding development and fabrication costs for the next-generation technology (which has already led to the formation of large international alliances in the semiconductor industry) (Isaac, 2003), in combination with more modular design strategies, will make it more likely that the industry seeks cost reduction by outsourcing its design and (possibly) fabrication work to foreign countries. Packaging and Integration Technologies In response to the realization that future enhancements of chip performance require major innova- tions and more heterogenous technology components, it is expected that the focus on integration and packaging will continue to increase, which is consistent with the view of the ITRS (see Figure 3-23). In Figure 3-23, monolithic system-on-chip (SoC) and polylithic system-in-package (SiP) technolo- gies are viewed to be critical to enabling the continued growth of information technology beyond Moore’s FIGURE 3-23  The minimum feature sizes of integrated circuit manufacturing have already passed the required scales for nanophotonic components, at least at telecommunication wavelengths, making the integration of photo- nics and electronics readily achievable using standard integrated circuit nanofabrication tooling. NOTE: CMOS, complementary metal oxide semiconductor; CPU, central processing unit; RF, radio frequency; HV, high voltage. 3-23 SOURCE: Semiconductor Industry Association (2005b). Reproduced with permission of SEMATECH.

ENABLING TECHNOLOGIES 121 law (i.e., “more Moore”). Figure 3-24 illustrates the balance between SoC and SiP solutions, which is naturally governed by the respective cost function. While certain system complexities and heterogeneous- ness can be integrated cost-effectively using a modular SoC approach, the integration of highly complex and heterogenous systems—possibly including nanophotonic devices, but also microelectromechanical systems (MEMS) and bioelectronics—is more economically addressed by SiP technologies with a higher degree of miniaturization and flexibility. Because nanophotonics is still an infant technology, it is arguably a very difficult task to under- stand which packaging and integration technology is most relevant for its enablement. The complicated interplay between limitations of existing technologies, the potential enablement of new computing a ­ rchitectures by optics and nanophotonics, the appropriate level of integration, and the advances of the different nanophotonic technology platforms is still an ongoing research topic. However, it is likely that the accessibility of nanophotonics will be significantly influenced by the way that computer manu­ facturers decide to integrate optics and nanophotonics (i.e., How close does optics have to get to the device and microprocessor to harness the benefits of optics cost-effectively?). The committee recommends the close monitoring of these trends in the integration of optics and nanophotonics by computer manufacturers because they will most likely determine the direction of massive investments. If indeed nanophotonics can be integrated cost-effectively and monolithically with high-performance CMOS, the respective suitable nanophotonic technology platforms will be more limited (Si-based and CMOS-compatible nanophotonics). Special attention has to be directed toward CMOS-compatible nanophotonics research and development activities because the nanophotonic tech- nology platform will have to be attuned with CMOS processes and materials. However, if optics will be integrated polylithically or heterogeneously, then the nanophotonic technology platforms can be much more diverse, and a corresponding technology watch would have to encompass a much wider range of activities. In the latter case, the role of packaging will be even more central. System on chip Time to market Cost / function Bio-interface Power supply MEMS ging pa cka 3D nd S IP a System complexity FIGURE 3-24  Cost versus complexity for system-on-chip, system-in-package (SiP), and 3D Packaging integration approaches. NOTE: MEMS, microelectromechanical systems; 3D, three dimensional. SOURCE: Semiconductor Industry Association (2005a). Reproduced with permission of SEMATECH. 3-24 redrawn

122 Nanophotonics The following subsections first discuss monolithic (i.e., Si photonics) and then polylithic/­heterogenous integration and packaging approaches (three-dimensional Si, Si-carrier). For purposes of discussion, the committee distinguishes between monolithic and polylithic (heterogeneous) integration, but it realizes that this boundary is somewhat blurred with the introduction of new integration and packaging tech­ nologies such as three-dimensional silicon. Monolithic Integration: Silicon Photonics Monolithic integration using silicon photonics drives toward the seamless fabrication of CMOS and photonic circuits on the same circuit layer, creating an SoC-like chip with an on-chip communica- tion layer (Reed and Knights, 2004). The United States is a leader in this field, driven by its dominant semiconductor industry, but there are also significant and technically relevant activities in Europe (in particular, at IMEC in Belgium, COM [Communications, Optics, and Materials] Research in Denmark, RWTH Aachen University, Germany); in Japan (at NTT); and also more recently in China, Korea, and India. The promise of Si photonics lies in the exploitation of the maturity and precision of CMOS technology as well as in the synergy of ultracompact integration of optical and electrical functions on the same chip. Silicon is transparent in the range of optical telecommunication wavelengths (at 1.55 micrometers [µm]) and has a high refractive index that allows for the fabrication of densely packed nanophotonic structures. The general concept of Si as a waveguide and of modulating its index using electrons and holes had been proposed for quite some time; only recently, however, has signifi- cant progress been made, in which CMOS-compatible waveguides (and other passive devices such as splitters), ­modulators, deflection switches, wavelength demultiplexers, delay lines, and detectors have been demonstrated (Bogaerts et al., 2005; Soref and Bennett, 1987; Soref and Lorenzo, 1986). In addition, complete chip-level ­demonstrations have been also been reported, which have implemented monolithic integration between Si photonics and CMOS using a standard CMOS fabrication line (Gunn, 2006; Jalali et al., 2005). While these demonstrations are clearly major stepping-stones, the monolithic integration between “high-performance” CMOS in the form of a microprocessor or appli- cation-specific circuits has not been done yet—partially because of integration issues (as discussed in more detail below) but also because the required optical components are not available yet. Waveguides and Passives The recent progress in silicon-based waveguides has been helped by the general availability of silicon on insulator (SOI) wafers, which can be readily used to fabricate Si waveguides supported by the buried oxide (BOX). SOI-based single mode (SM) and multimode (MM) waveguides have been fabricated with cross-sectional dimensions of approximately 0.3 × 0.3 µm2 and 0.3 × 2.0 µm2, respectively (Bogaerts et al., 2004; Vlasov and McNab, 2004). Despite the fact that propagation losses increase drastically with an increasing index of refraction contrast ratio (∆n = nSi – nSiO2 = 2), the unprecedented accuracy of CMOS fabrication capabilities (by fabricating waveguides with extremely small roughness) has yielded ultradense (pitch of less than 2.0 µm with negligent cross-talks) waveguide structures with quite low losses of less than 1.6 dB/cm and 0.2 dB/cm for SM and MM, respectively (Lee et al., 2001; Vlasov and McNab, 2004). Despite this progress, it remains an ongoing challenge because any losses directly impact the link budgets, which can be very tight in a power-limited design environment. Consequently, the optical properties of Si waveguides are still an active research topic, because propagation (as well as coupling and/or insertion losses) is dependent sensitively on group index, BOX thicknesses, exact wave-

ENABLING TECHNOLOGIES 123 guide dimensions, wavelengths, optical modes, distance to electrical contacts, and other interconnects. Following the successful fabrication of these waveguides, low-loss splitters, bends, directional couplers, and other passives such as wavelength demultiplexers and delay lines have been fabricated (Dumon et al., 2006; Xia et al., 2006; Xia et al., 2007). The dispersion properties of SM Si waveguides are shown to have a dispersion length of several centimeters at data rates of larger than 200 Gbps, which is suit- able for on-chip communications (Dulkeith et al., 2006). It is also established that these waveguides can carry up to 20 dBm of power, which is limited by nonlinear effects in the Si waveguides, fundamentally constraining the link budgets (Hsieh et al., 2006). Modulators Several recent demonstrations have shown the feasibility of compact modulators on SOI with up to 10 Gbps bandwidth (e.g., 6.5 Gbps and 1 milliwatt [mW] power dissipation) (Almeida et al., 2005). Fundamentally, there are two simple mechanisms (thermooptic and carrier injection) for modulating the refractive index in silicon, which can be exploited for building modulators. Modulators with both mechanisms have been demonstrated in several optical device configurations, such as Mach-Zehnder interferometers and ring resonators (Almeida et al., 2005; Espinola et al., 2003). The disadvantage with thermooptic modulators (typically realized in a Mach-Zehnder configuration) is the limited band- width, which is governed by the typically slow thermal diffusion times. While carrier-injection-based m ­ odulators can be very fast (limited by carrier lifetimes), they can suffer from poor overlap between the optical mode and the injected carriers, which require high Q resonator devices (to boost the sensitivity) and which are very temperature-sensitive. At this stage it remains a research topic to balance bandwidth, power consumption, losses, extinction ratio, temperature stability, and footprint requirements for these types of modulators. The committee notes that recently Si-based electrooptical modulators have also been demonstrated using quantum wells, although still with a low extinction ratio (Kuo et al., 2005). Detectors Compared with III/V detectors, Si-based receivers are lacking performance, although recently a very high bandwidth germanium-on-silicon-on-insulator (Ge-on-SOI) photodetector was demonstrated with a very small footprint (Dehlinger et al., 2004; Schow et al., 2006). CMOS-compatible detector design will be governed by a trade-off between CMOS compatibility (materials, processes, process flow), required bandwidth, wavelength range, sensitivity (quantum field, absorption depth), voltage requirements, and packaging approach. The detector will be closely integrated with the waveguide structure with minimum footprint to limit capacitance (Dehlinger et al., 2004; Schow et al., 2006). Other complementary detector approaches comprise tunnel junctions such as metal-insulator-metal, where nanoscopic optical antennae may be used to increase the coupling and detection efficiency (Zia et al., 2006). Light Sources or Gain Elements Despite intensive research, an Si laser or light source is still the central missing piece needed to enable the complete monolithic integration of Si photonics. As is well known, the reason for the difficul- ties in engineering a suitable light source lies in the fact that Si is an indirect band-gap material, which makes nonradiative processes more probable than the actual light emission. Heterogeneous approaches involve silicon evanescent amplifiers, which employ silicon waveguides and an evanescent tail extend- ing into indium gallium aluminum arsenide (InGaAlAs) quantum wells. These quantum wells, which

124 Nanophotonics are deposited by low-temperature, oxygen plasma-enhanced bonding, are very promising because these devices can be quite small, with low power consumption (Fang et al., 2006). The committee believes that nanophotonics will play an important role in the potential enablement of an Si-based light source and/or gain element. In the absence of an on-chip light source, light needs to be coupled on and off the chip, which is a technically difficult task owing to the high index of ­refraction contrast of the Si waveguides. However, significant progress has recently been made on two fronts. First, several groups managed to mode-match the different optical profiles between a large single-mode fiber and the Si waveguides by using a smart waveguide taper coupling the light adiabatically into the submicron waveguide (Day et al., 2003; Salib et al., 2004). Second, passive alignment techniques for the external optical fiber with respect to the waveguide have been demonstrated (by creating lithographi- cally defined structures on the silicon surface in order to align the fiber to the waveguide aperture); these techniques remove the need for a closed-loop optimization/coupling scheme. In summary, the emergence of Si photonics is likely to have a profound impact in large distance- c ­ ommunication systems. As Si photonics relates to the enablement of chip-scale supercomputers (enabling the unmatched communication between many cores), it is currently not clear how Si photonics will be integrated with the high-performance CMOS for many reasons and issues ranging from BOX thickness, metallization, circuit and device integration, and process compatibility. Heterogeneous Integration: Silicon Carrier, Three-Dimensional Silicon The monolithic integration of Si photonics with high-performance CMOS circuitry could be eventu- ally the most cost-effective integration way; it may turn out, however, that the optimization of ­electrical and optical components leads to such different competing requirements that only heterogeneous ­packaging approaches can rectify this. As discussed, these heterogeneous approaches will provide openings for more diverse technology platforms. Although the general trend to use Si as a package material somewhat favors Si photonic-based approaches (even if they are not entirely compatible with high-performance CMOS): specifically, an Si chip carrier can provide via densities, which make heterogeneous assembly approaches more feasible, where one needs very high-interconnect densities—formerly not supported by traditional carriers (Knickerbocker et al., 2005). The Si carrier technology components are manifold (Si through micro-C4, high-density wiring, and many supporting testing, assembly, and fabrication technologies) and have been demonstrated recently. In essence, an Si carrier provides a general way of very tightly integrating different chip technologies and achieving SoC performances with a system on a package (SoP) solution, thereby creating a “virtual” chip. This technology can play a key role in the enablement of nanophotonics—especially in conjunction with high-performance CMOS—without the burden of being “fully” CMOS-compatible. For example, a chip including nanophotonic devices (e.g., a receiver chip) can be readily integrated and packaged with a microcontroller or microprocessor using SoP technologies. Figure 3-25 illustrates a roadmap for this technology, with a trend toward increas- ing integration from two to three dimensions. It also shows that the Si carrier is the first important step toward even tighter three-dimensional integration, which is a trend consistent in the industry’s advancing chip-stacking techniques. Overarching recommendation Recommendation 3-1. The committee recommends that the intelligence technology warning com- munity be especially aware of countries and groups that have access to a combination of advanced packaging and integration technologies and nanophotonics.

ENABLING TECHNOLOGIES 125 Potential disruptive technology 3D integration Advantages Challenge I/O per cm2 3D Si integration and pkg 3D wiring Cooling 3D chip integration - I/O: 6- m pitch; 2.5M I/O/cm2 Net length Yield — 105 to 108 - Wiring pitch: 90 nm Performance Design Test Alignment Assembly Si carrier and chip stacking Si carrier pkg and through-via stacking - I/O: 50- m pitch; 40K I/O/cm2 Modular design Test Chip stack and Integration - Wiring pitch: 2 m Performance Assembly chip to Si carrier High bandwidth — 104 to 105 Chip 1 Chip 2 Chip mfg Substrate Organic and ceramic pkg (SCM and MCM) BGA, CGA, or LGA Ceramic and organic pkg Ceramic and organic pkg Existing pkg Limited Chip to pkg - I/O: 200- m pitch; 2.5K I/O/cm2 - I/O: 150- m pitch; 4K I/O/cm2 - I/O — 103 - Wiring pitch: MLC 200 m; - Wiring pitch: MLC 150 m; - Wiring SLC 50 m SLC 40 m - Bandwidth Printed circuit card Chip to board — 102 2000 2005 2010 2015 Year FIGURE 3-25  Silicon integration and packaging roadmap. NOTE: SCM, single-chip module; MCM, multichip module; Si, silicon; 3D, three-dimensional; I/O, input/output; BGA, ball grid array; MLC, multilayer ceramic; CGA, column grid array; LGA, land grid array; SLC, surface laminar circuit. SOURCE: Knickerbocker et al. (2005). Reproduced with permission of IBM Research. ������������������������������������������� References Almeida, Vilson R., Qianfan Xu, and Michal Lipson. ������������������������������������������������������������������������ 2005. Ultrafast integrated semiconductor optical modulator based on the plasma-dispersion effect. Optics Letters 30(18):2403-2405. Amon, Cristina H., Sartaj S. Ghai, Woo Tae Kim, and Myung S. Jhon. 2006. Modeling of nanoscale transport phenomena: Application to information technology. Physica A: Statistical Mechanics and Its Applications 362(1):36-41. Bängtsson, Erik, and Maya Neytcheva. 2005. Algebraic preconditioning versus direct solvers for dense linear systems as arising in crack propagation problems. Communications in Numerical Methods in Engineering 21(2):73-81. Benner, A.F., M. Ignatowski, J.A. Kash, D.M. Kuchta, and M.B. Ritter. 2005. Exploitation of optical interconnects in future server architectures. IBM Journal of Research and Development 49(4/5):755-775. Bernstein, K., D.J. Frank, A.E. Gattiker, W. Haensch, B.L. Ji, S.R. Nassif, E.J. Nowak, D.J. Pearson, and N.J. Rohrer. 2006. High-performance CMOS variability in the 65-nm regime and beyond. IBM Journal of Research and Development 50(4/5):433-449. Bita, Ion, Taeyi Choi, Michael E. Walsh, Henry I. Smith, and Edwin L. Thomas. 2007. Large-area 3D nanostructures with octagonal quasicrystalline symmetry via phase-mask lithography. Advanced Materials 19(10):1403-1407. Blanco, A., K. Busch, M. Deubel, C. Enkrich, G. von Freymann, M. Hermatschweiler, W. Koch, S. Linden, D.C. Meisel, G.A. Ozin, S. Pereira, C.M. Soukoulis, N. Tétreault, and M. Wegener. 2004. Three-dimensional lithography of photonic crystals. In Advances in Solid State Physics, edited by B. Kramer. ������������������������������������� Heidelberg, Germany: Springer Berlin.

126 Nanophotonics Bockstaller, Michael R., Rafal A. Mickiewicz, and Edwin L. Thomas. 2005. Block copolymer nanocomposites: Perspectives ��������������������������������������������������� for tailored functional materials. Advanced Materials 17(11):1331-1349. Bogaerts, W., D. Taillaert, B. Luyssaert, P. Dumon, J. Van Campenhout, P. Bienstman, D. Van Thourhout, R. Baets, V. Wiaux, and S. Beckx. 2004. Basic structures for photonic integrated circuits in silicon-on-insulator. Optics Express 12(8):1583‑1591. Bogaerts, W., R. Baets, P. Dumon, V. Wiaux, S. Beckx, D. Taillaert, B. Luyssaert, J. Van Campenhout, P. Bienstman, and D. Van Thourhout. 2005. Nanophotonic waveguides in silicon-on-insulator fabricated with CMOS technology. Journal of Lightwave Technology 23(1):401-412. Brueck, S.R.J. 2005. Optical and interferometric lithography-nanotechnology enablers. Proceedings of IEEE 93(10):1704‑1723. Cai, L.Z., X.L. Yang, and Y.R. Wang. 2002. Formation of three-dimensional periodic microstructures by interference of four noncoplanar beams. Journal of the Optical Society of America A 19(11):2238-2244. Campbell, M., D.N. Sharp, M.T. Harrison, R.G. Denning, and A.J. Turberfield. 2000. Fabrication of photonic crystals for the visible spectrum by holographic lithography. Nature 404(6773):53-56. Cao, Wenyi, A. Muñoz, P. Palffy-Muhoray, and B. Taheri. 2002. Lasing in a three-dimensional photonic crystal of liquid crystal blue phase II. Nature Materials 1(2):111-113. Chao, Weilun, Bruce D. Harteneck, J. Alexander Liddle, Erik H. Anderson, and David T. Attwood. 2005. Soft x-ray microscopy at a spatial resolution better than 15 nm. Nature 435(7046):1210-1213. Chen, H.H., J. Lee, J. Weiner, Y.K. Chen, and J.T. Chen. 2006. A 14-b 150 MS/s CMOS DAC with digital background calibra- tion. Paper read at the Symposium on VLSI Circuits, June15-17, 2006, Honolulu, Hawaii. Cheng, Joy Y., Anne M. Mayes, and Caroline A. Ross. 2004. Nanostructure engineering by templated self-assembly of block copolymers. Nature Materials 3(11):823-828. Cheng, Hongwei, William Y. Crutchfield, Zydrunas Gimbutas, Leslie F. Greengard, J. Frank Ethridge, Jingfang Huang, V ­ ladimir Rokhlin, Norman Yarvin, and Junsheng Zhao. 2006a. A wideband fast multipole method for the Helmholtz equation in three dimensions. Journal of Computational Physics 216(1):300-325. Cheng, Joy Y., Caroline A. Ross, Henry I. Smith, and Edwin L. Thomas. 2006b. Templated self-assembly of block copolymers: Top-down helps bottom-up. Advanced Materials 18(19):2505-2521. Chew, W.C., E. Michielssen, J.M. Song, and J.M. Jin. 2001. Fast and Efficient Algorithms in Computational Electromagnetics. Norwood, Mass.: Artech House. Chou, Stephen Y., Peter R. Krauss, Wei Zhang, Lingjie Guo, and Lei Zhuang. 1997. Sub-10 nm imprint lithography and a ­ pplications. Paper presented at the 41st International Conference on Electron, Ion, and Photon Beam Technology and Nanofabrication, Dana Point, Calif. Coles, H.J., M.J. Clarke, S.M. Morris, B.J. Broughton, and A.E. Blatch. 2006. Strong flexoelectric behavior in bimesogenic liquid crystals. Journal of Applied Physics 99(3):034104. Contopanagos, H., B. Dembart, M. Epton, J.J. Ottusch, V. Rokhlin, J.L. Visher, and S.M. Wandzura. 2002. Well-conditioned boundary integral equations for three-dimensional electromagnetic scattering. IEEE Transactions on Antennas and Propa- gation 50(12):1824-1830. Cumpston, Brian H., Sundaravel P. Ananthavel, Stephen Barlow, Daniel L. Dyer, Jeffrey E. Ehrlich, Lael L. Erskine, Ahmed A. Heikal, Stephen M. Kuebler, I.Y. Sandy Lee, Dianne McCord-Maughon, Jinqui Qin, Harald Rockel, Mariacristina Rumi, Xiang-Li Wu, Seth R. Marder, and Joseph W. Perry. 1999. Two-photon polymerization initiators for three-dimensional optical data storage and microfabrication. Nature 398(6722):51-54. Darve, E. 2000. The fast multipole method: Numerical implementation. Journal of Computational Physics 160(1):195-240. Day, I., I. Evans, A. Knights, F. Hopper, S. Roberts, J. Johnston, S. Day, J. Luff, H. Tsang, and M. Asghari. 2003. Tapered silicon waveguides for low insertion loss highly-efficient high-speed electronic variable optical attenuators. Paper read at IEEE Optical Fiber Communications Conference, March 23-28, 2003, Atlanta, Georgia. Dehlinger, G., S.J. Koester, J.D. Schaub, J.O. Chu, Q.C. Ouyang, and A. Grill. 2004. High-speed germanium-on-SOI lateral PIN photodiodes. IEEE Photonics Technology Letters 16(11):2547-2549. Dieringer, Jon A., Adam D. McFarland, Nilam C. Shah, Douglas A. Stuart, Alyson V. Whitney, Chanda R. Yonzon, Matthew A. Young, Xiaoyu Zhang, and Richard P. Van Duyne. 2006. Surface enhanced Raman spectroscopy: New materials, concepts, characterization tools, and applications. Faraday Discussions 132:9-26. Draine, Bruce T., and Piotr J. Flatau. 1994. Discrete-dipole approximation for scattering calculations. Journal of the Optical Society of America A 11(4):1491-1499. Dulkeith, Eric, Fengnian Xia, Laurent Schares, William M. Green, Lidija Sekaric, and Yurii A. Vlasov. 2006. Group index and group velocity dispersion in silicon-on-insulator photonic wires. Optics Express 14(9):3853-3863.

ENABLING TECHNOLOGIES 127 Dumon, P., W. Bogaerts, D. Van Thourhout, D. Taillaert, R. Baets, J. Wouters, S. Beckx, and P. Jaenen. 2006. Compact wavelength router based on a silicon-on-insulator arrayed waveguide grating pigtailed to a fiber array. Optics Express 14(2):664-669. Edrington, A.C., A.M. Urbas, P. DeRege, C.X. Chen, T.M. Swager, N. Hadjichristidis, M. Xenidou, L.J. Fetters, J.D. J ­ oannopoulos, Y. Fink, and E.L. Thomas. 2001. Polymer-based photonic crystals. Advanced Materials 13(6):421-425. Espinola, R.L., M.C. Tsai, J.T. Yardley, and R.M. Osgood, Jr. 2003. Fast and low-power thermooptic switch on thin silicon- on-insulator. IEEE Photonics Technology Letters 15(10):1366-1368. Fang, A.W., Park Hyundai, R. Jones, O. Cohen, M.J. Paniccia, and J.E. Bowers. 2006. A continuous-wave hybrid AlGaInAs- silicon evanescent laser. IEEE Photonics Technology Letters 18(10):1143-1145. Finkelmann, H., S.T. Kim, A. Muñoz, P. Palffy-Muhoray, and B. Taheri. 2001. Tunable mirrorless lasing in cholesteric liquid crystalline elastomers. Advanced Materials 13(14):1069-1072. Foresi, J.S., P.R. Villeneuve, J. Ferrera, E.R. Thoen, G. Steinmeyer, S. Fan, J.D. Joannopoulos, L.C. Kimerling, H.I. Smith, and E.P. Ippen. 1997. Photonic-bandgap microcavities in optical waveguides. Nature 390(6656):143-145. Frank, D.J. 2002. Power constrained CMOS scaling limits. IBM Journal of Research and Development 46(2/3):235-244. Gaillot, Davy P., Elton Graugnard, Jeffrey S. King, and Christopher J. Summers. 2007. Tunable Bragg peak response in liquid- crystal-infiltrated photonic crystals. Journal of the Optical Society of America B 24(4):990-996. Gaylord, T.K., and M.G. Moharam. 1985. Analysis and applications of optical diffraction by gratings. Proceedings of the IEEE 73(5):894-937. Greengard, L., Huang Jingfang, V. Rokhlin, and S. Wandzura. 1998. Accelerating fast multipole methods for the Helmholtz equation at low frequencies. IEEE Computing in Science and Engineering 5(3):32-38. Gunn, Cary. 2006. CMOS photonics for high-speed interconnects. IEEE Micro 26(2):58-66. Gunn, Cary. 2007. CMOS Photonics SOI Learns a New Trick! (Much More than Moore) [PDF of PowerPoint Presentation]. Luxtera Corporation. Available from http://www.luxtera.com/assets/conference_20060914_Soitec.pdf. Last accessed May 17, 2007. Guo, L. Jay. 2004. Recent progress in nanoimprint technology and its applications. Journal of Physics D: Applied Physics 37(11):R123-R141. Hersee, S.D., X. Sun, and X. Wang. 2006. The controlled growth of GaN nanowires. Nano Letters 6(8):1808. Ho, K.M., C.T. Chan, C.M. Soukoulis, R. Biswas, and M. Sigalas. 1994. Photonic bandgaps in 3-dimensions—New layer-by- layer periodic structures. Solid State Communications 89:413-416. Hori, T., J. Takayanagi, N. Nishizawa, and T. Goto. 2003. Flatly broadened, wideband and low noise supercontinuum genera- tion in highly nonlinear hybrid fiber. Optics Express 12(2):317-324. Horowitz, Mark. 2007. Scaling, power and the future of CMOS. Pp. 6-15 in the 20th International Conference on VLSI ­Design Held Jointly with 6th International Conference on Embedded Systems, Hyderabad, India. Available online at http:// csdl2.computer.org/persagen/DLAbsToc.jsp?resourcePath=/dl/proceedings/&toc=comp/proceedings/vlsid/2007/2762/00/ 2762toc.xml. Hsieh, I-Wei, Xiaogang Chen, Jerry I. Dadap, Nicolae C. Panoiu, Richard M. Osgood, Sharee J. McNab, and Yurii A. Vlasov. 2006. Ultrafast-pulse self-phase modulation and third-order dispersion in Si photonic wire-waveguides. Optics Express 14(25):12380-12387. Hu, Jiangtao, Liang-shi Li, Weidong Yang, Liberato Manna, Lin-wang Wang, and A. Paul Alivisatos. 2001. Linearly polarized emission from colloidal semiconductor quantum rods. Science 292(5524):2060-2063. Isaac, Randy. 2003. Perspectives on the semiconductor industry. Paper presented at The Future of the U.S. Semiconductor Industry Conference, May 8, 2003, Washington, D.C., IBM. Jalali, B., O. Boyraz, D. Dimitropoulos, and v. Raghunathan. 2005. Scaling laws of nonlinear silicon nanophotonics. Proceed- ings of the SPIE 5730:41-49. Kagan, C.R., T.L. Breen. and L.L. Kosbar. 2001. Patterning organic-inorganic thin-film transistors using microcontact printed templates. Applied Physics Letters 79(21):3536-3538. Kahle, J.A., M.N. Day, H.P. Hofstee, C.R. Johns, T.R. Maeurer, and D. Shippy. 2005. Introduction to the cell multiprocessor. IBM Journal of Research and Development 49(4/5):589-604. Kalra, Yogita, and R.K. Sinha. 2005. Design of ultra compact polarization splitter based on the complete photonic band gap. Optical and Quantum Electronics 37(9):889-895. Kashiwa, T., Y. Sendo, K. Taguchi, T. Ohtani, and Y. Kanai. 2003. Phase velocity errors of the nonstandard FDTD method and comparison with other high-accuracy FDTD methods. IEEE Transactions on Magnetics 39(4):2125-2128. Kato, Jun-ichi, Nobuyuki Takeyasu, Yoshihiro Adachi, Hong-Bo Sun, and Satoshi Kawata. 2005. Multiple-spot parallel process- ing for laser micronanofabrication. Applied Physics Letters 86(4):044102-3.

128 Nanophotonics Knickerbocker, J.U., P.S. Andry, L.P. Buchwalter, A. Deutsch, R.R. Horton, K.A. Jenkins, Y.H. Kwark, G. McVicker, C.S. Patel, R.J. Polastre, C. Schuster, A. Sharma, S.M. Sri-Jayantha, C.W. Surovic, C.K. Tsang, B.C. Webb, S.L. Wright, S.R. McKnight, E.J. Sprogis, and B. Dang. 2005. Development of next-generation system-on-package (SOP) technology based on silicon carriers with fine-pitch chip interconnection. IBM Journal of Research and Development 49(4/5):725-753. Knight, J.C., T.A. Birks, P.S.J. Russell, and D.M. Atkin. 1996. All-silica single-mode optical fiber with photonic crystal c ­ ladding. Optics Letters 21(19):1547-1549. Kumagai, J. 2001. Fighting in the streets [on high-tech military equipment]. IEEE Spectrum 38(2):68-71. Kuo, Yu-Hsuan, Yong Kyu Lee, Yangsi Ge, Shen Ren, Jonathan E. Roth, Theodore I. Kamins, David A.B. Miller, and James S. Harris. 2005. Strong quantum-confined Stark effect in germanium quantum-well structures on silicon. Nature 437(7063):1334-1336. Lange, Birger, Rudolf Zentel, Shalin J. Jhaveri, and Christopher K. Ober. 2006. 3D defect engineering in polymer opals. Paper presented at Photonic Crystal Materials and Devices III, April 3-6, 2006, Strasbourg, France. Lazzi, G., and O.P. Gandhi. ����������������������������������������������������������������������������������������������� 2000. A mixed FDTD-integral equation approach for on-site safety assessment in complex electro­ magnetic environments. IEEE Transactions on Antennas and Propagation 48(12):1830-1836. Lee, K.K., D.R. Lim, L.C. Kimerling, J. Shin, and F. Cerrina. 2001. Fabrication of ultralow-loss Si/SiO 2 waveguides by rough- ness reduction. Optics Letters 26(23):1888-1890. Li, Z.Y., I. El-Kady, K.M. Ho, S.Y. Lin, and J.G. Fleming. 2003. Photonic band gap effect in layer-by-layer metallic photonic crystals. Journal of Applied Physics 93(1):38-42. Lim, Jong H., Kyung S. Lee, Jin C. Kim, and Byeong H. Lee. 2004. Tunable fiber gratings fabricated in photonic crystal fiber by use of mechanical pressure. Optics Letters 29(4):331-333. Lin, S.Y., and J.G. Fleming. 1999. A three-dimensional optical photonic crystal. Journal of Lightwave Technology 17(11):1944‑1947. Lin, S.Y., J.G. Fleming, D.L. Hetherington, B.K. Smith, R. Biswas, K.M. Ho, M.M. Sigalas, W. Zubrzycki, S.R. Kurtz, and J. Bur. 1998. A three-dimensional photonic crystal operating at infrared wavelengths. Nature 394(6690):251-253. Liz-Marzan, L.M., M. Giersig, and P. Mulvaney. 1996. Synthesis of nanosized gold-silica core-shell particles. Langmuir 12(18):4329-4335. Lopez, C. 2003. Materials aspects of photonic crystals. Advanced Materials 15(20):1679-1704. Lu, Wei, and Charles M. Lieber. ������������������������������� Journal of Physics D: Applied Physics 39:R387-R406. 2006. Semiconductor nanowires. Manna, Liberato, Erik C. Scher, and A. Paul Alivisatos. 2002. Shape control of colloidal semiconductor nanocrystals. Journal of Cluster Science 13(4):521-532. Massoud, Y., and J. White. 2002a. FastMag: A 3-D magnetostatic inductance extraction program for structures with permeable materials. Proceedings of the 2002 IEEE/ACM International Conference on Computer-Aided Design, November 10-14, 2002, San Jose, Calif. Massoud, Y., and J. White. 2002b. Improving the generality of the fictitious magnetic charge approach to computing inductances in the presence of permeable materials. Proceedings of the 39th Design Automation Conference, June 10-14, 2002, New Orleans, La. Matsuhara, M., M. Kamura, and A. Maruta. 1991. Boundary element analysis of electromagnetic field in cylindrical structures. ������������������������������������������������������������������������������������ Paper read at International Conference on Computation in Electromagnetics, November 25-27, 1991, London, United Kingdom. Minhas, B.K., W. Fan, K. Agi, S.R.J. Brueck, and K.J. Malloy. 2002. Metallic inductive and capacitive grids: Theory and experiment. Journal of the Optical Society of America A 19(7):1352-1359. Mishchenko, Michael I., Larry D. Travis, and Andrew A. Lac. 2002. Scattering, Absorption, and Emission of Light by Small Particles. Cambridge, United Kingdom: Cambridge University Press. Moreira, M.F., I.C.S. Carvalho, W. Cao, C. Bailey, B. Taheri, and P. Palffy-Muhoray. 2004. Cholesteric liquid-crystal laser as an optic fiber-based temperature sensor. Applied Physics Letters 85(14):2691-2693. Norris, D.J., and Y.A. Vlasov. 2001. Chemical approaches to three-dimensional semiconductor photonic crystals. Advanced Materials 13(6):371-376. Oldenburg, S.J., R.D. Averitt, S.L. Westcott, and N.J. Halas. 1998. Nanoengineering of optical resonances Chemical Physics Letters 288(2):243-247. Palffy-Muhoray, Peter, Wenyi Cao, Michele Moreira, Bahman Taheri, and Antonio Munoz. 2006. Photonics and lasing in liquid crystal materials. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 364(1847):2747-2761. Park, Cheolmin, Jongseung Yoon, and Edwin L. Thomas. 2003. Enabling nanotechnology with self assembled block copolymer patterns. Polymer 44(22):6725-6760.

ENABLING TECHNOLOGIES 129 Pendry, J.B. and P.M. Bell. 1996. Transfer matrix techniques for electromagnetic waves. Photonic Band Gap Materials 315:203-228. Prodan, E., C. Radloff, N.J. Halas, and P. Nordlander. 2003. A hybridization model for the plasmon response of complex nanostructures. Science 302(5644):419-422. Reed, Graham T., and Andrew P. Knights. 2004. Silicon Photonics: An Introduction. West Sussex, England: Wiley. Reese, C., C. Becher, A. Imamoglu, E. Hu, B.D. Gerardot, and P.M. Petroff. 2001. Photonic crystal microcavities with self- a ­ ssembled InAs quantum dots as active emitters. Applied Physics Letters 78(16):2279-2281. Russell, Philip. 2003. Photonic crystal fibers. Science 299(5605):358-362. Sakurai, T., and Y. Watanabe. 2000. Advances in Scanning Probe Microscopy. Heidelberg, Germany: Springer. Salib, Mike, Ling Liao, Richard Jones, Mike Morse, Ansheng Liu, Dean Samara-Rubio, Drew Alduino, and Mario Paniccia. 2004. Silicon photonics. Intel Technology Journal 8(2):143-160. Schow, C.L., L. Schares, S.J. Koester, G. Dehlinger, R. John, and F.E. Doany. 2006. A 15-Gb/s 2.4-V optical receiver using a Ge-on-SOI photodiode and a CMOS IC. IEEE Photonics Technology Letters 18(19):1981-1983. Schroden, R.C., M. Al-Daous, C.F. Blanford, and A. Stein. 2002. Optical properties of inverse opal photonic crystals. Chemistry of Materials 14(8):3305-3315. Semiconductor Industry Association. 2005a. International Technology Roadmap for Semiconductors, 2005 edition. Austin, Tex: SEMATECH. Available online at http://www.itrs.net/Links/2005ITRS/AP2005.pdf Semiconductor Industry Association. 2005b. International Technology Roadmap for Semiconductors, 2005 edition. Austin, Tex: SEMATECH. Available online at http://www.itrs.net/Links/2005ITRS/ExecSum2005.pdf. Shevchenko, Elena V., Dmitri V. Talapin, Nicholas A. Kotov, Stephen O’Brien, and Christopher B. Murray. 2006. Structural diversity in binary nanoparticle superlattices. Nature 439(7072):55-59. Shoji, S., H.B. Sun, and S. Kawata. 2003. Photofabrication of wood-pile three-dimensional photonic crystals using four-beam laser interference. Applied Physics Letters 83(4):608-610. Sohler, Wolfgang, Bijoy K. Das, Dibyendu Dey, Selim Reza, Hubertus Suche, and Raimund Ricken. 2005. Erbium-doped lithium niobate waveguide lasers. IEICE Transactions on Electronics E88-C(5):990-997. Soref, R.A., and J.P. Lorenzo. 1986. All-silicon active and passive guided-wave components for λ = 1.3 and 1.6 microns. IEEE Journal of Quantum Electronics QE-22:873-879. Soref, Richard A., and Brian R. Bennett. 1987. Electrooptical effects in silicon. IEEE Journal of Quantum Electronics QE- 23:123-129. Stöber, W., A. Fink, and E. Bohn. 1968. Controlled growth of monodisperse silica spheres in the micron size range. Journal of Colloid Interface Science 26:62-69. Stolk, P.A., F.P. Widdershoven, and D.B.M. Klaassen. 1998. Modeling statistical dopant fluctuations in MOS transistors. IEEE Transactions on Electron Devices 45(9):1960-1971. Sullivan, D., Liu Jun, and M. Kuzyk. 2000. Three-dimensional optical pulse simulation using the FDTD method. IEEE Trans- actions on Microwave Theory and Techniques 48(7):1127-1133. Sun, Hong-Bo, Shigeki Matsuo, and Hiroaki Misawa. 1999. Three-dimensional photonic crystal structures achieved with two- photon-absorption photopolymerization of resin. Applied Physics Letters 74(6):786-788. Taflove, A. 1980. Application of the finite-difference time-domain method to sinusoidal steady-state electromagnetic-­penetration problems. IEEE Transactions on Electromagnetic Compatibility EMC-22(3):191-202. Tang, Zhiyong, Zhenli Zhang, Ying Wang, Sharon C. Glotzer, and Nicholas A. Kotov. 2006. Self-assembly of CdTe nanocrystals into free-floating sheets. Science 314(5797):274-278. Taotao, Lu, Wang Zeyi, and Yu Wenjian. 2004. Hierarchical block boundary-element method (HBBEM): A fast field solver for 3-D capacitance extraction. IEEE Transactions on Microwave Theory and Techniques 52(1):10-19. Ullal, C.K., M. Maldovan, M. Wohlgomuth, and E.L. Thomas. 2003. Triply periodic bicontinuous structures through ­interference lithography: A level set approach. Journal of the Optical Society of America A 20(5):948-954. Urbas, A., M. Maldovan, P. DeRege, and E.L. Thomas. 2002. Bicontinuous cubic block copolymer photonic crystals. Advanced Materials 14(24):1850-1853. Urbas, A., J. Klosterman, V. Tondiglia, L. Natarjan, R. Sutherland, O. Tsutsumi, T. Ikeda, and T.J. Bunning. 2004. Optically switchable Bragg reflectors. Advanced Materials 16(16):1453-1456. Van den hove, Luc. 2002. Advanced Lithography. Presentation to IEEE International Electron Devices, December 9, 2002, San Francisco, Calif: IMEC. van Eijkelenborg, Martijn A., Alexander Argyros, Geoff Barton, Ian M. Bassett, Matthew Fellew, Geoffrey Henry, Nader A. Issa, Maryanne C.J. Large, Steven Manos, Whayne Padden, Leon Poladian, and Joseph Zagari. 2003. Recent progress in microstructured polymer optical fibre fabrication and characterisation. Optical Fiber Technology 9(4):199-209.

130 Nanophotonics Vlasov, Yurii, and Sharee McNab. 2004. Losses in single-mode silicon-on-insulator strip waveguides and bends. Optics Express 12(8):1622-1631. Wang, Y. 1991. Nonlinear optical properties of nanometer sized semiconductor clusters. Accounts of Chemical Research 24:133-139. Wang, Shumin, and F.L. Teixiera. 2004. Some remarks on the stability of time-domain electromagnetic simulations. IEEE Transactions on Antennas and Propagation 52(3):895-898. Wang, X., J.F. Xu, H.M. Shu, Z.H. Zeng, Y.L. Chen, H.Z. Wang, Y.K. Pang, and W.Y. Tam. 2003. Three-dimensional photonic crystals fabricated by visible light holographic lithography. Applied Physics Letters 82(14):2212-2214. Wang, G.T., A.A. Talin, D.J. Werder, J.R. Creighton, E. Lai, R.J. Anderson, and I. Arslan. 2006a. Highly aligned, template- free growth and characterization of vertical GaN nanowires on sapphire by metalorganic chemical vapour deposition. Nanotechnology 17(23):5773-5780. Wang, H., D.W. Brandl, F. Le, P. Nordlander, and N.J. Halas. 2006b. Nanorice: A hybrid plasmonic nanostructure. Nano ­Letters 6(4):827-832. Wendt, J.R., G.A. Vawter, P.L. Gourley, T.M. Brennan, and B.E. Hammons. 1993. Nanofabrication of photonic lattice structures in GaAs/AlGaAs. Journal of Vacuum Science and Technology B 11(6):2637-2640. Wiersma, Diederik S., and Stefano Cavalieri. 2001. Light emission: A temperature-tunable random laser. Nature 414(6865):708‑709. Williams, Benjamin S., Sushil Kumar, Qi Qin, Qing Hu, and John L. Reno. 2006. Terahertz quantum cascade lasers with double-resonant-phonon depopulation. Applied Physics Letters 88(26):261101. Wolchover, N.A., F. Luan, A.K. George, J.C. Knight, and F.G. Omenetto. 2007. High nonlinearity glass photonic crystal nanowires. Optics Express 15(3):829-833. Xia, Y.N., B. Gates, and Z.Y. Li. 2001. Self-assembly approaches to three-dimensional photonic crystals. Advanced Materials 13(6):409-413. Xia, Fengnian, Lidija Sekaric, Martin O’Boyle, and Yurii Vlasov. 2006. Coupled resonator optical waveguides based on silicon- on-insulator photonic wires. Applied Physics Letters 89(4):041122. Xia, Fengnian, Lidija Sekaric, and Yurii Vlasov. 2007. Ultracompact optical buffers on a silicon chip. Nature Photonics 1(1):65-71. Xiang, Jie, Wei Lu, Yongjie Hu, Yue Wu, Hao Yan, and Charles M. Lieber. 2006. Ge/Si nanowire heterostructures as high- performance field-effect transistors. Nature 441(7092):489-493. Yablonovitch, E., and T.J. Gmitter. 1989. Photonic band structure: The face-centered-cubic case. Physical Review Letters 63(18):1950-1954. Yamamoto, Y. 2006. Quantum communication and information processing with quantum dots. Quantum Information Process- ing 5(5):299-311. Yarotski, Dzmitry A., and Antoinette J. Taylor. 2004. Ultrafast scanning tunneling microscopy: Principles and applications. Pp. 57-98 in Ultrafast Dynamical Processes in Semiconductors, edited by K.-T. Tsen. Heidelberg, Germany: Springer Berlin. Yee, Kane S. 1966. Numerical solution of initial boundary value problems involving Maxwell’s equations in isotropic media. IEEE Transactions on Antennas and Propagation 14(3):302-307. Yen, Brian. 2006. Microfluidic Reactors for the Synthesis of Nanocrystals. Dissertation, Department of Chemistry, Massachu- setts Institute of Technology, Cambridge, Mass. Yen, Brian K.H., Axel Günther, Martin A. Schmidt, Klavs F. Jensen, and Moungi G. Bawendi. 2005. A microfabricated gas- liquid segmented flow reactor for high-temperature synthesis: The case of CdSe quantum dots. Angewandte Chemie 44(34):5447-5451. Yin, Yadong, Robert M. Rioux, Can K. Erdonmez, Steven Hughes, Gabor A. Somorjai, and A. Paul Alivisatos. 2004. Formation of hollow nanocrystals through the nanoscale Kirkendall effect. Science 304(5671):711-714. Yoon, J., W. Lee, and E.L. Thomas. 2006. Optically pumped surface-emitting lasing using self-assembled block-copolymer- distributed Bragg reflectors. Nano Letters 6(10):2211-2214. Young-Seek, Chung, Cheon Changyul, Son Joo-Hiuk, and Hahn Song-Yop. 2000. FDTD analysis of propagation characteristics of terahertz electromagnetic pulses. IEEE Transactions on Magnetics 36(4):951-955. Yu, Y.Y., S.S. Chang, C.L. Lee, and C.R.C. Wang. 1997. Gold nanorods: Electrochemical synthesis and optical properties. Journal of Physical Chemistry B 101(34):6661-6664. Zhong, Y.C., S.A. Zhu, and H.Z. Wang. 2005. Fabrication of compound lattice by holographic lithography. Chinese Physics Letters 22(2):369-372. Zia, Rashid, Jon A. Schuller, and Mark L. Brongersma. 2006. Near-field characterization of guided polariton propagation and cutoff in surface plasmon waveguides. Physical Review B (Condensed Matter and Materials Physics) 74(16):165415.

Next: 4 Potential Military Applications of Nanophotonics »
Nanophotonics: Accessibility and Applicability Get This Book
×
Buy Paperback | $69.00 Buy Ebook | $54.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The Committee on Technology Insight-Gauge, Evaluate & Review set up by the NRC at the request of the Defense Intelligence Agency, has selected a number of emerging technologies to investigate for their potential threats to and opportunities for national security. This first study focused on emerging applications of nanophotonics, which is about the interaction of matter and light at the scale of the wavelength of the light. Manipulation of matter at that scale allows tailoring the optical properties to permit a wide-range of commercial and defense applications. This book presents a review of the nanoscale phenomena underpinning nanophotonics, an assessment of enabling technologies for developing new applications, an examination of potential military applications, and an assessment of foreign investment capabilities

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!