2.1. Two-photon absorption
Multiphoton absorption (MPA) was predicted as early as 1931  and experimentally observed immediately after the invention of lasers . Two-photon absorption (TPA) can be conceptualized as follows: molecules exposed to high-intensity light can undergo near simultaneous absorption of two photons mediated by a “virtual state”. The combined energy of the two photons accesses an excited state of the molecule. TPA thus involves the interaction of both photons that combine their energies to produce an electronic excitation analogous to that conventionally caused by a single photon of a correspondingly shorter wavelength. Unlike single-photon absorption, whose probability is linearly proportional to the incident intensity, the TPA process depends on both a spatial and temporal overlap of the incident photons and takes on a quadratic (nonlinear) dependence on the incident intensity, resulting in highly localized excitation with a focused light beam. In the early 1990s, some potential applications with this technique were demonstrated, such as two-photon fluorescence (TPF) 3D data storage  and TPF microscopy . However, the TPA efficiency of the materials was an issue for these applications, and this has triggered the effort to search for new materials with higher TPA efficiency, see, e.g., the works of the groups of Prasad [62-64] and Marder [65-67], as well as the work of other groups [68-70] for the design of better TPA materials. TPA remains a very active field of research, and recently, with the advent of femtosecond lasers, multiphoton excitation processes found applications in fluorescence imaging of biological molecules [71-75], fluorescence upconversion and two-photon lasing [76, 77], power limiting , photodynamic therapy [79-81], optical memory by rewritable photoisomerization and photorefraction effect [60, 82-90], cell surgery [91, 92], synchronization of embryonic heart beating and stimulation of cardiomyocite cells [93, 94], and micro-nanofabrication including micro-optical components, photonic crystals, and complicated 3D patterning by photopolymerization [49, 95-136].
2.2. Photopolymerization and two-photon materials
Photopolymerization is a photoreaction that converts a liquid or gel monomer into a solid polymer. These reactions require the use of a photosensitive molecule, e.g. photoinitiator, that produces an active species upon light irradiation. Most monomers and oligomers commonly do not produce an initiating species for polymerization, so it is necessary to introduce low molecular weight organic compounds that act as photoinitiators. An additional requirement for effective initiation of a photopolymerization process is a photosensitizer that is capable of absorbing light and transferring the excitation energy to the photoinitiator. An ideal photoinitiator should be easily synthesized, low priced, nontoxic, odorless, highly shelf stable (high thermal stability and stability in darkness), and have a high absorption of incident light and a high quantum yield in the generation of the active moieties, and highly soluble in the polymerization medium. The wavelength of the irradiation light can be in the UV, visible or infrared region. When a photoinitiator absorbs a suitable photon energy, it is transformed into a reactive initiating species like radicals or cation radicals. Cationic initiators are photoacid generators that produce cations upon light irradiation. The latter are used for the polymerization of epoxides or vinyl ethers via a cationic polymerization mechanism. The photolysis of cationic photoinitiators has been shown to result in the generation of free radicals allowing the combination of monomers to be polymerized. Conventional photopolymerization initiators generate free radicals upon light irradiation, which initiate a free-radical polymerization process of acrylates or vinyl ethers. The most commonly used free-radical photoinitiator is benzophenone and its derivatives. Free-radical polymerizations are chain reactions in which the addition of a monomer molecule to an active chain-end regenerates the active site at the chain-end. The free-radical photopolymerization mechanism involves at least three different kinds of reactions [137-139]: The first one is the initiation during which the free-radical initiator is decomposed with light in the presence of a monomer to form an active species. In the next reaction, known as the propagation, the initiator fragment reacts with a monomer molecule to form the first active adduct that is capable of being polymerized. Monomers continue to add in the same manner resulting in the formation of macroradicals that are end-active polymers. The final reaction is the termination, during which the growth center is deactivated and the final polymer molecules are formed. This step normally involves the reaction between two polymers bearing active centers and can proceed by two different mechanisms, combination or disproportionation, leading to the formation of one or two polymer chains, respectively:
Here I, I*, M, R• denote a photoinitiator, an intermediate state of I after absorbing a photon, a monomer, and a radical, respectively. hν refers to the energy of an absorbed photon. Besides the above, other reactions, such as chain transfer and chain inhibition, often take place and complicate the mechanism of free-radical polymerization. Photopolymerizations follow the general scheme for any polymerization, however, the use of light, rather than heat, to drive the reaction has certain advantages, such as the elimination of solvent, the high reaction rates at room temperature and the spatial control of the polymerization. Photopolymerization differs from photocrosslinking, which involves the absorption of a photon in every chain propagation step. Thus, the quantum yield of photopolymerization per absorbed photon can reach thousands.
It is beyond the scope of this paper to present an exhaustive review of the materials and molecules used in two-photon polymerization, a recent comprehensive summary of this topic can be found elsewhere . In this paper, we will briefly, review the main classes of two-photon materials. The first materials employed in two-photon polymerization were acrylic photopolymers and the negative photoresist SU8. Over the last few years, two-photon polymerization research has focused on photosensitive sol-gel hybrid materials  such as the commercially available ORMOCER (ORganically-MOdified-CERamic, MicroResist) [142-144]. The sol-gel process is based on the phase transformation of a sol of metallic oxide or alkoxide precursors to form a wet gel. A photosensitive sol-gel process usually involves the catalytic hydrolysis of the sol-gel precursor(s) and the polycondensation of the hydrolyzed products and other sol-gel-active components present in the reaction medium to form a macromolecular hybrid network structure. The gel formed is subsequently reacted through photopolymerization to give a product similar to glass.
Silicate-based photopolymers have proved to be a very popular choice in microfabrication using multiphoton polymerization [142, 145-147]. They combine the properties of silicate glasses such as hardness, chemical and thermal stability, and optical transparency with the laser processing at low temperatures of organic polymers, properties impossible to achieve with just inorganic or polymeric materials. The most widely used hybrid material is ORMOCER. This material comprises an inorganic (–Si–O–Si–) backbone that can be functionalized with a range of organic functionalities. While ORMOCER and other silicate-only based hybrid materials have provided the possibility to fabricate high-resolution 3D structures with good optical properties, they do not allow the optimization and “fine tuning” of the material properties for specific applications. The versatile chemistry of sol-gel composites allows the copolymerization of more than one metal alkoxide, this has been shown to enhance the material's mechanical stability and allows the modification of its optical properties [148-152]. There are several examples of 3D microstructures fabricated using two-photon polymerization, and acrylate photopolymers doped with metal or metal-oxide nanoparticles, e.g. composite materials [153, 154]. However, despite the obvious advantages of preparing true composite photosensitive sol-gels, there are very few such examples in two-photon polymerization. Active hybrid materials are also the subject of intense research because they add functionality to the fabricated structures. Such functionalities may include photochemistry and photoswiching, nonlinearity, conductivity, and so on.
2.3. Two-photon photopolymerization
3D pholithography, e.g. 3D stereolithography by femtosecond lasers, is performed by scanning a tightly focused laser beam into a photopolymerizable resin from the bottom surface to the top surface to fabricate an entire 3D structure (negative photoresist). See, e.g., Fig. 1. In this work, e.g. that corresponding to Fig. 1, a titanium sapphire laser operating in mode-lock at 76 MHz and 780 nm with a 150-fs pulse width was used as an exposure source and it was focused by an objective lens of high numerical aperture (∼1.4) yielding an intensity on the sample in the order of 1013 W/μm2. This is a typical intensity for fs-laser fabrication. Two- or multiphoton absorption occurs at the high photon density of a laser beam. The wavelength of the fs-laser light is in the near-infrared (NIR) region, a range in which most of the photoresists are transparent, a feature that allows a deep penetration in the medium with negligible absorptive power loss. Nonlinear absorption, e.g. TPA or MPA and photoreaction occur in a volume, e.g. voxel, much smaller than the cube of the wavelength, λ3, of the laser, a feature that is a key to high special resolution [105, 106]. The precision of the focal spot sample positioning is obtained by a galvanomirror set in the plane of the sample and a piezostage in the out-of-plane direction. A charge-coupled device (CCD) camera is used for monitoring the focal spot on the sample and in situ fabrication .
Figure 1. (Top) Scanning electron microscopic (SEM) image of a microbull sculpture produced by raster scanning (left) and surface profile scanning (right). For the latter, only the crust was defined by the two-photon process, the inside was solidified by illumination under a mercury lamp. In this particular structure, the two-photon scanning time was reduced by 90% due to the use of surface profile scanning and subsequent solidification under a mercury lamp. Nanofabrication took 13 min. (Bottom) schematics of (a) raster, and (b) surface profile scanning. Reproduced with permission from .
Download figure to PowerPoint
Stereo photolithography dates back to the work of Kodama in 1981  and it was further developed by several other researchers to improve the speed, and the resolution [156-163]. However, the dimensions of the fabricated structures are several orders of magnitudes larger than those that can be fabricated by TPA or MPA processes. Kawata's group developed the technology of fs 3D nanofabrication as reported in a series of works (see for example, references 49 and 95). The absorption of UV-curable resin is due to valence electron transition and it is located in the UV or the blue side of the visible spectrum, and when exposed to NIR, TPA or MPA occurs. Evidence of the occurrence of nonlinear absorption can be obtained when using a mode-locked femtosecond laser. Linear absorption is governed by photon flux, (photon/cm2), expressed by the product of irradiation duration and photon flux intensity (photon/s cm2), which is proportional to light intensity (W/cm2). The adequate energy dose needed for absorption is reachable either by intense irradiation for short time or weak irradiation for long times. To date there is no experimental report on polymerization observed when NIR lasers work in CW mode, where the power is uniformly delivered over time. On the other hand, polymerization occurs with the same irradiation dose under pulsed mode. When switched to pulsed mode, 82 MHz, 100 fs and 1 mW mean power, a diffraction-limited focal spot has a photon flux density of 1029 photons/s cm2 in the pulse duration, which is 1.2×105 times larger than when the laser is operating in CW mode. This shows that polymerization is due to a nonlinear optical absorption effect.
The first reported real 3D microstructure  is a 7-μm diameter and 50-μm long spiral coil with a line cross section of 1.3–2.2 μm (Fig. 2). Due to the use of a relatively low NA (0.85) objective, the feature size is larger than the potential limit provided by the technology. This work confirmed the feasibility of two-photon photopolymerization in three dimensions. In this experiment, a 790-nm, 200-fs laser was focused into SCR 500 [Japan Synthesis Rubber Company, JSR], a commercial urethane acrylate resin. Both nonlinear optical absorption and thresholding, e.g. chemical and material photoreactive nonlinearity (vide infra), contribute to high spatial resolution of nanofabrication. Nonlinear absorption results from a light-intensity distribution that is spatially narrower than that of linear absorption used for UV lithography . For example, the squared light intensity distribution related to TPA reduces the volume of interaction of light with matter, thereby improving the resolution of nanofabrication. Thresholding is a nonlinear photochemical effect. For example, for two-photon induced polymerization of resins, the photogenerated radicals are spatially distributed, obeying the square law of the light intensity distribution. The rate of photopolymerization of monomers is therefore expected to follow the same distribution. However, due to the existence of the radical quenchers, like dissolved oxygen molecules, radicals survive and initiate polymerization only at the region with highest light intensity, and thus a threshold is formed. Since the light intensity relative to the threshold is continuously adjustable by exposure duration or laser pulse intensity, the size of a voxel, which defines the spatial resolution, can be tuned to be much smaller than that defined by the diffraction limit [110-114]. This effect is called material nonlinearity .
Figure 2. Scanning electron microscopic (SEM) image of a spiral coil structure made by two-photon photopolymerization after removal of unsolidified resin. (a) A view of the entire structure and (b) the magnified top end. Reproduced with permission from .
Download figure to PowerPoint
Nanofabricated structures may shrink in two-photon lithography, an effect that can be precompensated in the computer design of the structure , or used to advantage to reduce the size of the fabricated structure . There are also resins that do not shrink after nanofabrication. One of which is SU-8, which is an ultrathick negative photoresist that has much more strength than liquid-based resins [166, 167]. Figure 3 shows fabricated 3D spiral structures without shrinkage employing two-photon-induced crosslinking in SU-8.
Figure 3. (a) Schematic principle of the fabrication of a spiral structure in SU-8 by direct laser writing. SEM images of spiral-type photonic crystals fabricated with a pulse energy of 0.6 nJ, (b) full shape that has a period of 1.8 μm, length of spiral arm (L) of 2.7 μm, and a vertical pitch (P) of 3.04 μm as illustrated in (c). (d) and (e), Fabricated circular-spiral structure with a L-shaped waveguide and its magnified image, respectively. Reprinted with permission from .
Download figure to PowerPoint
Besides laser scanning, laser interference has been used in the fabrication of optical gratings, 2-dimensional rod arrays, 3-dimensional diamond-like photonic crystals and 3-dimensional networks [168-172]. An interesting example of nanofabrication of functional nanostructures is shown in Fig. 4. Indeed, this figure shows 2D and 3D patterns of CdS-polymer nanocomposites fabricated by a four-beam interference technique in films with various thicknesses . Micro-nanofabrication by laser-beam interference and single-beam irradiation is discussed in Section 'Nanofabrication in azopolymers' for azopolymers.
Figure 4. SEM images of microstructures of CdS–polymer nanocomposites fabricated by four-beam laser interference with various film thicknesses and exposure times, (a) thickness, 2 μm, exposure time, 20 s, (b) thickness, 2 μm, exposure time, 10 s, (c) thickness, 2 μm, exposure time, 8 s, (d) thickness, 10 μm, exposure time, 30 s, (e) thickness, 10 μm, exposure time, 20 s, (f) thickness, 10 μm, exposure time, 10 s. The experimental details can be found in . The scale bar is 2 μm in all images. Reproduced with permission from .
Download figure to PowerPoint
2.4. Photopolymerization beyond the diffraction limit of light
We will go on to discuss more applications of two-photon photopolymerization. Innovative techniques of multiphoton polymerization that are based on stimulated-emission-depletion (STED) fluorescence microscopy, developed by Hell and Wichmann  and Hell's group , have been recently introduced. They include single-photon rather than multiphoton excitation , a one-color scheme , a multiphoton two-color scheme [178, 179], and diffusion-assisted STED . In STED, fluorescent molecules are excited by a short laser pulse. A second laser pulse, which is tuned to a substantially longer wavelength than the first pulse, is used to de-excite the molecules through stimulated emission. This depletion pulse must arrive after vibrational relaxation is complete in the excited electronic state but before significant fluorescence has occurred. Spatial phase shaping of the depletion beam causes de-excitation to occur everywhere except in a region at the center of the original focal volume. The size of this region depends on the intensity of the depletion beam and the corresponding degree of saturation of stimulated emission. Thus, fluorescence can be localized in a zone much smaller than the excitation wavelength. World-record lateral resolutions down to 5.6 nm using visible light and nitrogen-vacancy centers in diamond have been reported by Hell's group . It is interesting to translate this success of optical microscopy to optical lithography. This possibility was already mentioned by Hell's group in 2000 .
In principle, STED should work equally well to deactivate polymerization in multiphoton polymerization or conventional photopolymerization. Typical radical photoinitiators undergo intersystem crossing to a triplet state on a time scale on the order of 100 ps . The radicals that lead to polymerization are formed in the triplet state, so de-excitation of molecules before intersystem crossing occurs will turn off photopolymerization. Furthermore, radical photopolymerization only occurs above a threshold concentration of radicals, so de-excitation of a small fraction of excited photoinitiator molecules could be sufficient to halt polymerization if the concentration of excited molecules is just above this threshold. Li et al.  achieved λ/20 resolution by one-color initiation and deactivation of photopolymerization. They used a technique inspired by STED that they call resolution augmentation through photoinduced deactivation (RAPID) lithography. In RAPID lithography, one laser beam is used to initiate polymerization in a negative-tone photoresist. A second laser beam is used to deactivate the photoinitiator, preventing photopolymerization from occurring. Multiphoton absorption of pulsed 800 nm light is used to initiate crosslinking in a polymer photoresist and one-photon absorption of continuous-wave (CW) 800-nm light is used simultaneously to deactivate the photopolymerization. By spatial shaping of the phase of the deactivation beam, features far smaller than the excitation wavelength can be fabricated. They demonstrated the fabrication of features with scalable resolution along the beam axis, down to a 40-nm minimum feature size.
Scott et al.  demonstrated a technique of photopolymerization, which is also inspired by STED, and that uses a two-color irradiation scheme, whereby initiating species are generated by single-photon absorption at one wavelength (473 nm from a CW laser) while inhibiting species are generated by single-photon absorption at a second, independent wavelength (364 nm from a CW Argon ion laser). Coirradiation at the second wavelength thus reduces the polymerization rate, delaying gelation of the material and facilitating enhanced spatial control over the polymerization. Appropriate overlapping of the two beams produces structures with both feature sizes and monomer conversions otherwise unobtainable with the use of single- or two-photon absorption photopolymerization. Additionally, the generated inhibiting species rapidly recombine when irradiation with the second wavelength ceases, allowing for fast sequential exposures not limited by memory effects in the material and thus enabling fabrication of complex two- or three-dimensional structures (Fig. 5) . This figure demonstrates 65-nm dots with a single-photon fabrication process. The optical approach demonstrated in  produces confinement of the polymerized region along only two axes, and manipulation of the photoinhibiting wavelength into a bottle beam profile  would induce confinement along the third axis, thus allowing fabrication of 3D structures with sub-100-nm isotropic resolution. Because single-photon absorption cross sections are often orders of magnitude larger than two-photon cross sections, this photoinitiation–photoinhibition system facilitates the use of inexpensive CW diode lasers and very high write velocities. Thus, this single-photon approach to nanolithography uses dramatically cheaper hardware and scales to much higher throughput.
Figure 5. Scanning electron micrographs of polymerized features. (a) Voxels polymerized on a microscope slide using a 0.45-NA singlet lens and a coincident Gaussian blue / UV irradiation scheme. The blue power was held constant at 10 μW while the UV was progressively increased. The UV power, from left to right, was 0, 1, 2.5, 10, and 100 μW. The exposure time was 8 s for each dot. Scale bars, 10 μm. (b) Profile of a voxel similarly fabricated but with 10 μW of blue power and 110 μW of UV focused at 1.3 NA, then imaged via SEM at normal incidence. The SEM intensity on the white line (inset) is plotted as squares against the expected polymerization profile (green) obtained by a double-parameter fit of the initiation (blue) and inhibition (violet) rate profiles, as shown. (c) Polymer column fabricated by using the same conditions as Fig. B. The focus was translated normal to the glass slide at a velocity of 0.125 μm/s for 3 μm. The high aspect ratio caused the column to fold during the solvent wash, leaving it lying on the glass surface. Scale bar, 200 nm. Reproduced with permission from .
Download figure to PowerPoint
Along the same lines, e.g. a two-color photopolymerization scheme based on STED, Gu's group fabricated 40 nm features . They used a direct laser writing technique based on single-photon photoinhibited polymerization. They used a CW-mode 375-nm inhibiting laser for inhibiting polymerization and a CW mode 488-nm initiating laser. The super-resolution feature is realized by overlapping two laser beams of different wavelengths to enable the wavelength-controlled activation of photoinitiating and photoinhibiting processes in the polymerization.
Fisher and Wegener  explored stimulated emission as a superior depletion mechanism and developed a multiphoton two-color photopolymerization scheme based on STED. The strategy of this scheme is briefly described as follows. They use a pulsed 810-nm Ti-Sapphire fs laser for photoinitiation by two-photon absorption and a cw green (532 nm) laser as a depletion beam. The relevant states and transitions of a photoinitiator molecule are schematically shown in Fig. 6a. The excitation (or writing) laser excites photoinitiator molecules from their ground states to an electronically and vibrationally excited level S1* via two-photon absorption. Some molecules may directly decay back to the ground-state nonradiatively. However, the majority of molecules will relax to an intermediate state S1 from where they can either fluoresce or undergo intersystem crossing (ISC) to the triplet state T1. From the T1, a chemical reaction may be initiated in the photoresist. The idea of STED is to bring the molecules from the intermediate state S1 back to the ground state S0 via stimulated emission (SE) induced by a second laser of a different color that is called the depletion laser.
Figure 6. (a) Energy-level scheme with transitions in a photoinitiator molecule for stimulated emission- depletion (STED) direct-laser-writing optical lithography. SE, stimulated emission, ESA, excited-state absorption, TTA, triplet–triplet absorption, ISC, intersystem crossing. (b) Ingredients of the photoresist used by the authors, pentaerythritol tetraacrylate (monomer) and 7-diethylamino-3-thenoylcoumarin (photoinitiator). Reproduced with permission from .
Download figure to PowerPoint
From this scheme, it can be seen that the search of photoinitiators is a key to the success of STED direct laser writing. The photoinitiator selected by these authors is shown in Fig. 6b. These authors succeeded in fabricating a photonic crystal woodpile structure with a clear stopgap by using STED direct laser writing with 52 layers (corresponding to a total height of 6.4 μm) and a rod spacing a = 350 nm, which is the intralayer periodicity (Fig. 7a). The quality of the sample interior was excellent over its entire thickness. Figure 7b shows the corresponding optical spectra with a pronounced stop band and a minimum transmittance of 3.5%.
Figure 7. (a) Oblique-view electron micrograph of a woodpile photonic crystal with 52 layers and a rod spacing of a = 350 nm made by STED direct laser writing. The sample has been milled with a focused-ion beam to reveal its interior. (b) Corresponding reflectance and transmitted spectra (normalized to substrate transmittance and the reflectance of an 80-nm silver film, respectively). Reproduced with permission from .
Download figure to PowerPoint
The implementation of direct laser writing with STED is complicated, requiring very fine beam control and photoinitiators that have not only a high two-photon cross section but also high fluorescence quantum efficiency [176-179]. When absorbing the light, the photoinitiator becomes excited and produces radicals that attack monomer molecules to form macroradicals of the polymer. Finally, the radicals are terminated to give the polymer. Radical termination can be also induced by oxygen and other molecules in the system, known as scavengers. Quenching competes with photopolymerization and is usually considered detrimental to the process. In multiphoton photopolymerization, however, it can be used to circumvent the diffraction limit and produce structures of very high resolution. Farsari's group  showed that it is possible to increase the resolution of multiphoton polymerization by employing a scheme based on quencher diffusion, in a chemical equivalent of STED. This is based on the combination of a mobile quenching molecule and a slow laser scanning speed, allowing the diffusion of the quencher in the scanned area, the depletion of the generated radicals, and the regeneration of the consumed quencher. They fabricated woodpile structures with an interlayer period of 400 nm, comparable to what has been achieved by STED direct laser writing, as discussed in the previous paragraph.
2.5. Applications of two-photon photopolymerization
Two-photon photopolymerization has found applications in photonic crystals and metamaterials fabrications [185-193] as well as in biomedical applications, including biodegradable materials and protein microstructuring [194-198]. We do not intend to review all the literature in these kinds of applications, rather we will highlight some important achievements in application of two-photon fabrication. For example, Wegener group  fabricated 3D woodpile photonic crystals with defects, e.g. they created a simple resonator by altering just one elementary cell of the periodic structure (Fig. 8). As the woodpile photonic crystal is constructed from a two-rod motif, they actually leave out exactly one of these building blocks to create a defect. Farsari's group  reported three-dimensional metallic photonic crystals with optical bandgaps in the near-IR to optical region. They prepared woodpile structures and they metalized them with silver. The fabricated structures had 900-nm and 600-nm periodicity and resolution below 100 nm. Using sol-gel chemistry, direct laser writing and electroless plating metallization techniques, they were able to create the appropriate polymer network, fabricate the structures and metalize them selectively. Farsari's group  also reported 3D structuring and metallization of a zirconium-based organic–inorganic photosensitive material doped with metal-binding tertiary amine moieties.
Figure 8. (a) Schematic representation of the resonator design after silicon double-inversion. (b) Focused ion beam cross section of a silicon photonic crystal with defect. (c) Numerical calculation of the mode profile. The gray structure represents the distribution of the silicon and acts as a guide for the eyes. (d) Perpendicular incidence transmittance spectra into the forward direction (entire half-space) of a structure containing one resonator, green (blue) solid curve. The inset is a close-up of the transmittance in the forward direction. The gray area displays the width of the complete photonic bandgap. Reproduced with permission from .
Download figure to PowerPoint
Metal nanostructures have been fabricated by Prasad's group  by two-photon lithography. They fabricated highly conductive gold nanostructures within a polymeric matrix with subwavelength resolution. The nanostructures are directly written in a gold-precursor-doped photoresist using a femtosecond pulsed laser. The laser energy is absorbed by a two-photon dye, which induces simultaneous reduction of gold in the precursor and polymerization of the negative photoresist. This results in gold-nanoparticle-doped polymeric lines that exhibit both plasmonic effects, due to the constituent gold nanoparticles, and relatively high conductivity due to the high density of particles within these lines. Line widths from 150 to 1000 nm have been achieved with this method, and several functional structures have been written including optically active planar chiral structures and plasmonic nanostructures, which are used in the development of composite metamaterials. Gu's group  fabricated and characterized three-dimensional photonic structures that represent a novel class of biomimetic photonic chiral composites inspired by butterfly wings. They fabricated different types of chiral composites that can be engineered to form novel photonic devices. They experimentally showed strong circular dichroism and compare with numerical simulations to illustrate the high quality of their 3D photonic crystals.
Metamaterials are artificial materials that, unlike naturally occurring substances, exhibit magnetism at optical frequencies, it is therefore essential that the structures are conducting. While most studies on fs-laser fabrication have focused on the realization of polymeric structures [49, 95-116], this technique was recently used for the fabrication of metallic 3D patterns over large areas using a microlens array [199-202], a feature which is a step forward from standard, single-beam laser writing into a polymer matrix, that is time consuming and not suitable for large-scale fabrication. The proposed method, however, enabled the simultaneously writing of more than 700 polymer structures that were uniform in size. The metallization of the structures was then achieved through the deposition of thin films composed of small silver particles by means of electroless plating (Fig. 9). A hydrophobic coating on the substrate prevented silver deposition in unwanted areas and allowed the formation of a large number of isolated and highly conducting objects. This method of metallization is flexible in that it can produce either polymer structures covered with metal or numerous isolated insulating polymer objects spread over a metallic film, depending on the resin properties and treatment procedures [201, 202]. Femtosecond-laser writing is considered to be one of the most promising methods for future manufacturing of large-area, true 3D metamaterials, offering intrinsic 3D parallel processing capability with good resolution (100 nm), and it can be successfully combined with selective metal deposition by electroless plating [201, 202].
Figure 9. Scanning electron microscope images of (a) self-standing empty cubic structures (height∼4.6 μm) connected in pairs and (b) a silver-coated polymer structure composed of a cube (2 μm in size) holding up a spring (inner diameter 700 nm). The structures are made by a two-photon-induced photopolymerization technique combined with electroless plating. Reprinted with permission from .
Download figure to PowerPoint
Recently, Pendry  pointed out an alternative route to negative refraction by exploiting chirality in combination with a resonant system with a photonic bandgap. Wegener's group  fabricated, by direct laser writing, chiral photonic crystals consisting of 3D dielectric (photoresist) helices arranged vertically on a two-dimensional (2D) square lattice. Transmittance measurements for circularly polarized light propagating along the helix axis found stop bands when the pitch of the dielectric helices matched the wavelength of light in the dielectric. Since this structure was uniaxial, its optical properties depended strongly on the angle of incidence. In a subsequent publication, Wegener's group  presented bichiral dielectric photonic crystals with a 3D simple cubic lattice. The chosen architecture was inspired by blue-phase cholesteric liquid crystals [206-208]. The photonic crystal consisted of three sets of 3D dielectric helices, one set oriented along each of the three orthogonal axes of the simple cubic lattice. Giessen's group  fabricated and characterized a metallic counterpart of the 3D bichiral dielectric crystals starting from the same point as Wegener's group. Giessen's group first fabricated a 3D bichiral crystal by two-photon femtosecond-laser writing in a negative-tone photoresist. Next, this dielectric template was coated with a conformal metal film via electroless silver plating. Electroless plating is a wet-chemical metallization technique based on an autocatalytic redox reaction that does not require an external current source, and thus coated all exposed surfaces. Finally, since electroless plating also coated the substrate with silver, the silvered crystal was detached with a glass capillary and transferred to a clean substrate in order to facilitate transmission spectroscopy. Note that standard deposition methods such as sputter coating or vacuum evaporation are unsuitable for metal coating due to self-shadowing of the complex 3D geometry.
Along the same lines of the work of Kawata's group [201, 202], Wegener's group  fabricated a uniaxial gold helix metamaterial shown in Fig. 10 (Left). Such structures can only be fabricated by direct laser writing. The polymer structures produced through direct laser writing lithography can be filled with gold using electroplating  (Fig. 10 (Right). Electroplating setups can be extremely simple and inexpensive, often requiring only a bias voltage between a transparent electrode on the substrate and a macroscopic counterelectrode within a beaker. The metamaterial sample footprint and height is therefore limited only by the direct laser-writing lithography process itself. Past achievements and future challenges in the development of three-dimensional photonic metamaterials have reviewed by Soukoulis and Wegener .
Figure 10. (Left) Chiral metamaterial made using direct laser writing and electroplating. (Right) Connected cubic-symmetry negative-index metamaterial structure amenable to direct laser writing . Reproduced with permission from . In the schematics the yellow and blue colors represent gold and dielectric materials, respectively.
Download figure to PowerPoint
Farsari's group  fabricated a functional photonic crystal, e.g. a photonic crystal made of a nonlinear optical material that consists of photosensitive sol-gel material that contains the nonlinear optical chromophore Disperse Red 1, an azodye chromophore, and that can be structured accurately using the two-photon polymerization technique. Farsari's group used this material to fabricate 3D photonic crystals with band stops in the near-infrared. The optical properties of such a functional photonic crystal can be controlled by the material's optical nonlinearity and photochemical switching capabilities of the azodye chromophore .
Another area where the potential of two-photon laser technology is being realized is in biomedical applications. Several photopolymers are biocompatible and able to be structured in complicated 3D shapes, a feature that makes them attractive for applications such as implantable microelectromachining systems (MEMS), and drug-delivery devices, such as microneedle arrays, devices that enable transdermal delivery of different pharmacological substances, and scaffolds for tissue engineering [97, 121, 124]. Figure 11 (left) shows an example of implantable MEMS, e.g. a microvalve designed to prohibit the reversal of blood flow in human veins, which may be caused by standing for too long. The valve is designed to open under forward fluid flow and close firmly in the case of backward flow. Figure 11 (right) shows an example of a high-porosity tissue engineering scaffold. The material used for two-photon fabrication of the valve and scaffold was a zirconium/silicon hybrid sol-gel SZ2080 . The scaffolds are required in order to produce living tissue that can integrate with host tissue inside the body. Micrometer-sized topography has an essential role in determining cell adhesion and surface-bound characteristics, influencing important cellular functions such as survival, proliferation, differentiation, and migration or mediator release. 3D cell cultures offer a more realistic local environment where the properties of cells can be observed and manipulated. An important factor in the production of tissue-engineering scaffolds is the ability to reproducibly manufacture 3D nanostructures, a feature that is a direct application of fs-laser fabrication scanning technology.
Figure 11. , Biomechanical parts made by two-photon fabrication. (left) Microvalve designed to prohibit the reversal of blood flow in human veins. Only part of the valve cover was built, to enable visualization of the interior. (Right) Scanning electron microscope image of high-porosity tissue engineering scaffold. The material used for two-photon fabrication was a zirconium/silicon hybrid sol-gel. Reproduced with permission from .
Download figure to PowerPoint
Engelhardt et al.  showed that two-photon polymerization offers the possibility of creating artificial cell scaffolds composed of micro- and nanostructures with spatial resolutions of less than 1 μm. For use in tissue engineering, the identification of a two-photon processable polymer that provides biocompatibility, biofunctionality and appropriate mechanical properties is a difficult task. Extracellular matrix proteins such as collagen or fibronectin, which could mimic native tissues best, often lack the mechanical stability. Hence, by generating polymer–protein hybrid structures, the beneficial properties of proteins can be combined with the advantageous characteristics of polymers, such as sufficient mechanical stability. They demonstrated the fabrication of 3D microstructures of polymer–protein hybrid structures and crosslinked gelatin microstructures, which clearly forced porcine chondrocytes to adapt their cell morphology, e.g. chondrocyte grows on crosslinked gelatin microlines, thereby showing the tissue engineering capabilities of two-photon fabrication. Along the same lines, Farsari's group  reported laser-induced photocrosslinking of protein microstructures. They fabricated 2D and 3D microstructures of avidin, bovine serum albumin and biotinylated bovine serum albumin. The multiphoton absorption-induced photocrosslinking of proteins was demonstrated with a nontoxic biomolecule flavin mononucleotide as the photosensitizer. Submicrometer- and micrometer-scale structures were fabricated from several different compositions of protein and photosensitizer. In addition, the retention of ligand-binding ability of the crosslinked protein structures was shown by fluorescence imaging of immobilized biotin or streptavidin conjugated fluorescence labels.
Seidlits et al.  reported high-resolution patterning of hydrogels in 3D using direct-write photofabrication for cell guidance. The development of 3D, spatially defined neuronal cultures that mimic chemical and physical attributes of native tissue is of considerable interest for various applications, including the development of tailored neuronal networks and clinical repair of damaged nerves. Seidlits et al. used multiphoton excitation to photocrosslink protein microstructures within 3D, optically transparent hydrogel materials, such as those based on hyaluronic acid. These methods can be used to create complex 3D architectures that provide both chemical and topographical cues for cell culture and guidance, providing a means to direct cell adhesion and migration on size scales relevant to in vivo environments. Using this approach, Seidlits et al. demonstrated guidance of both dorsal root ganglion cells and hippocampal neural progenitor cells along arbitrary, three-dimensional paths.
Two-photon fabrication of scaffolds for neural tissue engineering applications were reported by Farsari's group  . These authors reported the production of high-resolution 3D structures of polylactide-based materials via multiphoton polymerization and explored their use as neural-tissue engineering scaffolds. To achieve this, they synthesized a liquid polylactide resin and rendered it photocurable via attaching methacrylate groups to the hydroxyl end groups of the small molecular weight prepolymer. This resin cures easily under UV irradiation, using a mercury lamp, and under femtosecond IR irradiation. The results showed that the photocurable polylactide resin can be readily structured via direct laser writing with a femtosecond Ti:sapphire laser and submicrometer structures can be produced. The maximum resolution achieved is 800 nm. Neuroblastoma cells were grown on thin films of the cured polylactide material, and cell viability and proliferation assays revealed good biocompatibility of the material.
Sun's group  used two-photon fabrication to prepare protein microarchitectures. The protein molecules are crosslinked by a two-photon polymerization approach to form a hydrogel. They fabricated protein microlenses from bovine serum albumin by maskless femtosecond-laser direct writing. They used a commercial bovine serum albumin and a photosensitizer (methylene blue) to fabricate micro/nanoarchitectures. The photocrosslinked protein microstructures exhibited rapid and reversible swelling-to-shrinking behavior when stimulated by chemical signals. As a result, the focal distance of the as-formed microlenses can be tuned regularly and reversibly by changing the pH value. Because of their dynamically adjustable properties and full biocompatibility, microlenses show great promise for optical, electronic, and biomedical applications.
Both photoresists and azopolymers can be structured by light, but each with distinctive features and response to light absorption. Furthermore, two-photon direct laser writing is starting to be used for the combination of these two-kind of materials for added functionality . In contrast to photoresists, azopolymers have their own way of being structured by light, the reaction is not photopolymerization, it is rather induced mass movement of the polymer that fades away from light. This is what we discuss next.