1. INTRODUCTION

In recent years, there has been an incredible surge of technological advancement in additive manufacturing (3D printing). This has enabled rapid automated one-step fabrication of elements otherwise requiring multi-step processes by highly skilled specialists. This geometrical freedom allows great design flexibility, enabling new forms not possible with other techniques, as well as the combination of a variety of materials in the same process, permitting complex functionalities. Many of these technologies have made it to the consumer market and are widely used by start-ups and established companies in industries such as biomedical [1], automotive and aerospace [2], construction [3], food [4], and education [5]. These technologies include: (1) vat polymerization [6]; (2) material jetting [7,8], including 3D inkjet (chemical droplet bonding) and polyjet (UV-cured droplets of liquid photopolymer); (3) material extrusion, including fused deposition modeling (FDM) also termed fused filament fabrication (FFF) [9,10] and direct ink writing; and finally (4) two-photon printing, which is a vat polymerization method that allows for the highest resolution printing [11,12]. Among these methods, a variety of structures and applications are made possible owing to the range of print precisions and materials available.

3D printing has developed into a mature technology in a range of industries. However, 3D printing of optical and photonic components has had uneven progress among different technological approaches. For example, two-photon polymerization methods have had a strong presence in research for nearly two decades with commercial implementations available (e.g., Nanoscribe, Multiphoton Optics). Multiple applications and proof-of-concept devices were demonstrated through a sizeable body of publications [13–16]. On the other hand, many other printing techniques [e.g., selective laser sintering (SLS), FDM, jet-based printing, and stereolithography (SLA)] are still an emerging field with rapidly evolving technology. Only a few technologies have been applied in commercial product manufacturing. These include early examples of lens fabrication [17], light guide prototyping [18], and a number of special function components [19]. While multiple printing approaches have been commercialized, the field continues to grow rapidly, advancing to tackle issues in several key areas: (1) form precision and roughness (especially for large scale components) [7], (2) availability of optical materials [20], (3) limitations of the number of materials available in the same process [21], and (4) system versus component level prototyping [22]. Specifically, while it is possible to obtain high precision in a non-linear lithographic process, the manufactured parts are small and are constructed of one material, complicating corrections and limiting the range of possible geometries. On the other hand, technologies capable of larger scales and multiple materials lack the precision needed for optical quality surfaces and material density control. Resolving these challenges would create enormous new opportunities, such as (a) complex multi-component systems in a single process (also allowing tighter tolerancing and the ability to push system design limits), (b) arbitrary refractive geometries and new design concepts, (c) arbitrary light-guiding distributed imaging, (d) application-based customized device size, (e) new optical composite materials allowing design simplification, and (f) combining detection, actuation, and signaling in one development process.

ISO/ASTM 52900 standard terminology for printing methods includes material extrusion (which incorporates FFF/FDM printing and direct ink writing), vat polymerization (which includes stereolithography, digital light processing, and two-photon printing), and material jetting (which includes inkjet printing). In this review, separate focus will be given to FDM, direct ink writing, the vat polymerization techniques of stereolithography and digital light processing, and finally the vat polymerization technique of two-photon printing in order to adequately address the advantages and disadvantages of each specialized technique.

To discuss this, we will first describe the principles of the most common 3D printing technologies and present fabrication examples broken into common functional component classes (lenses–single and compound, arrays, waveguides, and other special components). We will also present a technology versus performance guide and an outline of available commercial systems.

In addition to the specialized additive manufacturing reports quoted throughout the text, there have been several other reviews written in relevant select areas of 3D printing. In order to distinguish this review from previously published reviews, the scope and new information of this review is provided, as well as the scope of prior reviews on adjacent and related topics. A practical aim of this work is to provide a review of additive manufacturing methods for optical fabrication through all scales in connection to practical limiting parameters like printing time, feature size and resolution, and surface roughness and form errors. This review has been structured so that specific types or components broadly used are discussed in their functional groups and point to specific manufacturing techniques that are appropriate within this group. Prior reviews focused on selected aspects of 3D printing technology. For example, Zhang et al. have presented a comprehensive review on glass 3D printing [23]. Jonušauskas et al. have done the same for methods of 3D printing that use optical systems, specifically with a focus on mesoscale printing [24]. An overview of laser nano-technology is given by Ye et al. [25]. Lin et al. provide a characterization of micro-additive manufacturing methods, specifically those which have been enabled by photopolymerization [26]. Blachowicz et al. [27] discuss the field of printing polymer optical elements and Camposeo et al. discuss additive manufacturing in photonics and optoelectronics [28]. While Zhang et al. [23], Jonušauskas et al. [24], Ye et al. [25], and Lin et al. [26] provide excellent reviews of the current state of 3D printing, their discussions are not focused on optical elements and optical system printing. Blachowicz et al. [27] is largely a survey of principles and does not go into quantitative comparisons of different approaches. In contrast, our review aims to go deeper into the analysis of the operation, performance, and design factors for 3D printing of various optical elements. Finally, Camposeo et al. [28] focuses on printing in connection to interconnect and optoelectronics. Here, we provide guidance on the design requirements for various optical elements and identify key optical performance metrics. In addition, we provide a more detailed accounting of efforts undertaken using material jetting techniques and material extrusion techniques such as FDM fabrication. In the context of this review, the print detail refers to the resolution of each printed voxel, or the smallest achievable feature size, and these terms are used interchangeably. While, generally, detail is limited by the printing mechanism, optimization processes have been used to achieve various feature sizes in each axis [9,29]. Print roughness refers to the surface quality of (and error in) the surfaces of the finished printed part, including post-processing steps. These values have a direct impact on achieved wavefront errors.

2. OVERVIEW OF 3D PRINTING TECHNOLOGIES

A graphical representation of a select number of 3D printing technologies is shown in Fig. 1. The graph presents Fused Deposition Modeling (FDM), Polyjet printing, Stereolithography (SLA), Direct Ink Writing (DIW), and Two-Photon Polymerization (2PP).

 

Fig. 1. Diagrams of the additive manufacturing methods.

Download Full Size | PDF

A. Fused Deposition Modeling

In FDM, the most common approach for consumer grade 3D printers, a thermoplastic filament is heated to its glass transition temperature, extruded through a nozzle, and deposited layer-by-layer [30]. The thermoplastic stiffens when cooled, forming the printed component. Suitable optical materials include transparent filaments Poly(methyl methacrylate) (PMMA), ColorFabb XT-clear (a commercial copolyester material comprised mostly of Polyethylene terephthalate glycol), Taulman3D Tech-G (a commercial Polyethylene terephthalate glycol material), Taulman3D t-glase (a commercial polyethylene cotrimethylene terephthalate), and Ultimaker CPE$+$ (a commercial polycyclohexylenedimethylene terephthalate glycol). Custom filaments can also be readily manufactured using filament extruders (e.g., Filbot) that take polymer pellets for raw material.

The resolution of the FDM printed components is primarily limited by the diameter of the extrusion nozzle, which is greater than 100 µm for commercial systems [31]. FDM is relatively inexpensive and can quickly produce macroscale designs, but this technique suffers from poor resolution and surface quality [32,33]. While this intrinsically limits its application in the production of optical components that require shape deviations and geometric tolerances on the nanometer scale, many functional optical components have been printed for the THz range [34–40]. At optical wavelengths, FDM has primarily been researched as a technique for fabricating fiber optic components [41–43].

B. Direct Ink Writing and Inkjet Printing

Direct Ink Writing (DIW) is a material extrusion technique in which a syringe or nozzle is mechanically translated with a 3D stage and deposits either a continuous filament or droplets of material that solidify layer by layer [44]. Ink materials include photopolymerizable polymers [20,45], ceramics [46–48], and colloidal suspensions [49]. A more advanced two-step writing and sintering process was demonstrated by Nguyen et al. in 2017 [50]. A colloidal suspension of silica was deposited from a 250 µm nozzle layer by layer and treated using a multi-step thermal processing technique. Lower temperatures are used to dry the material to a silica powder, and higher temperatures are used to consolidate the silica into a transparent glass. As in FDM, the main limitations in feature size are contained to the dimensions of the nozzle used for printing as well as the precision of the linear stages used to move the nozzle. Photopolymers have also been used as ink materials in the fabrication of optical waveguides [29,51,52]. Direct ink writing enables the fabrication of freeform geometries for waveguides with minimal post-processing steps. Currently, print detail and roughness limit the manufacture of high-precision optical components. So this technique is likely most suitable for illumination applications or imaging applications where microscopic spatial resolution is unnecessary.

Modern inkjet printing is a material jetting technique in which droplets of liquid photopolymer are deposited onto a substrate through one or more inkjet heads [8]. Resolution and feature sizes are limited by the material properties of the droplets and size of nozzles used to distribute the droplets. This technique has been researched as a tool for manufacturing fiber optics [53–56], waveguides for terahertz radiation [57,58], as well as lenses and microlens arrays [59–61]. Recently, macroscale inkjet-printed optical components have emerged on the market [62]. The company, Luxexcel, manufactures ophthalmic lenses and freeform optical components using a custom inkjet-printing technique, with a UV Photocurable polymer, Opticlear. Opticlear has a refractive index of 1.5225 [63], giving it similar properties to silica.

C. Vat Polymerization–SLA and DLP

Vat polymerization is an additive manufacturing process that uses UV-curable resin and focused UV light to build up parts in a layer-by-layer process. This method has a better feature resolution (25–100 µm) compared to extrusion-based filament printing. Transparent materials provided with commercial printers are usually proprietary and need to be characterized for optical properties. Some printers (e.g., Anycubic Photon Mono) allow filling of the printing reservoir with non-proprietary material. Vat polymerization works by either (1) laser sequential point-by-point (stereolithography) or (2) digital light processing (DLP) single-photon polymerization. Sequential point-by-point polymerization allows for smaller detail features, while DLP enables higher printing speeds. Post-print curing of the entire sample, which results in a highly homogeneous component, usually follows polymerization. This homogeneity makes vat polymerization a promising technique for lens fabrication, provided external surfaces are post-processed after fabrication to reduce surface roughness. As an example, spin-coating or curing against a glass surface have proven to be effective post-processing methods [64]. Additionally, these printers are relatively inexpensive through the consumer market (e.g., the Form3 or the Anycubic Photon Mono) and print with significantly faster speeds than two-photon based methods described below. The ability to print 3D optical elements with a consumer printer such as an SLA unit greatly expands the accessibility of optical element fabrication by keeping production costs and overall infrastructure low.

D. Two-Photon Printing

Two-photon printing (2PP) has emerged as an attractive option for optical fabrication because of the high resolution afforded by the multi-photon process. Unlike stereolithography, no post-processing is needed to achieve an optical quality surface. Most commonly, a femtosecond pulsed laser in the near-infrared region is used to polymerize a small region of photoresist around the focus. The two-photon absorption process limits the size of the polymerized focal volume. Hence, the high resolution (${\sim}{{150}}\;{\rm{nm}}$ minimum feature size). The optical component is formed by sequentially illuminating the photoresist, point-by-point, in a dense scanning pattern. As a consequence, 2PP requires long printing times, and thus is often limited to small print volumes. (Estimates for solid mode printing with commercial equipment (Nanoscribe, ${{25\times}}$ objective) are between 1 hr/mm³ and 320 hr/mm³, depending on the choice of printing resolution parameters. It is possible to print with higher speeds by utilizing a shell and scaffolding instead of solid printing but, for optics, this approach is often unusable.) This was recently somewhat mitigated by the implementation of laser power modulation, where higher laser power is used to increase the size of the polymerized focal volume (see, for example, Quantum X, Nanoscribe Inc.). Laser scanning is used to polymerize within the printer objective’s field of view (FOV). Larger parts require mechanical scanning and tiling of multiple FOVs. In order to avoid artifacts at the borders, significant optimization of the process is required [65]. Finally, a variety of polymeric photoresists are available, offering different mechanical and optical properties which could be exploited to manufacture more complex components.

Several custom-built 2PP systems have been developed to address the slow speed and small volume issues of the technology. Yang et al. used a spatial light modulator to create multiple focal spots parallelizing the print process, leading to reduced print times [66]. A similar approach was also demonstrated by Yan et al. and Tsutsumi et al. [67,68]. Instead of parallelizing the print process, Pearre et al. [69] implemented a resonant mirror for laser scanning and was able to generate speeds approaching 8000 mm/s. For comparison, typical printing speeds for two-photon systems range from 1 to 400 mm/s [69,70]. Stitching artifacts at the border between FOVs can severely compromise the functionality of the element. In order to address this issue, Jonušauskas et al. developed a method to use continuous writing by synchronizing the laser scanning and linear stage. Thus, avoiding the need for tiling and stitching artifacts [71].

Components with heights exceeding the working distance of the printer objectives are produced with dip-in laser lithography (DiLL), where the objective is dipped directly into the printing resin. Unfortunately, only a limited number of commercial objectives can work with the resin functioning as the immersion media. Obata et al. moved the cover glass, objective, and immersion oil together through the resist material to overcome this issue and produced components with heights as high as 7 mm [72]. An additional limitation of two-photon printing is aberrations in the polymerizing laser beam. In an effort to solve this, Hering et al. used a spatial light modulator to introduce automated aberration correction into lithography systems to improve fabrication resolution [73].

3. CLASSES AND APPLICATIONS OF PHOTONIC AND OPTICAL COMPONENTS AND SYSTEMS

The choice of 3D printing method is generally determined based on a compromise between print volume, speed, and resolution. For instance, 2PP techniques allow for high detail resolution but come at the expense of size limitations. Therefore, 2PP will most commonly be used for micro-optics and photonics applications like fiber couplers. Inkjet printers, on the other hand, enable larger size optics and are used in parallel custom lens printing at up to centimeter scales [7]. However, from a user perspective, it seems that the discussion of 3D printing techniques is more practical from the component level perspective rather than the specific technology group. Therefore, below we will discuss groups of most common optical components to suggest appropriate printing technology. This will include waveguides, fiber structures and couplers, lenses (single and compound) and lens arrays, and diffractive and phase modulation components.

A. Waveguide and Fiber Components

Optical waveguides are a foundational component in many micro-optical designs, enabling complex arrangements and piping of photons along arbitrary trajectories. Additive manufacturing offers several advantages for fabricating optical waveguides, including precise control of waveguide trajectories, dense packing of cores, and design freedom for waveguide geometry (e.g., photonic crystal fiber structures). Controlling waveguide performance, e.g., transmission loss, bend loss, and spatial modes, is very challenging for printing methods with resolutions comparable to, or larger than, typical core diameters in commercial optical fibers. In addition, printing methods achieving higher optical quality, such as 2PP, DIW, and inkjet, tend to be limited to centimeter waveguide lengths by print area and, thus, applications such as integrated optics, while FDM can be used to create kilometer length waveguides. Figure 2 shows examples and promising applications of 3D printed waveguides across optical (and neighboring) wavelengths created with different printing technologies.

 

Fig. 2. (a) Overview of the application wavelength and print detail for 3D-printed waveguides cited in this section. Superscripted numbers correspond to the cited work, while the print method is color-coded. (b)–(e) Illustrate promising aspects of the 3D printing methods toward application. (b) Inkjet printing can be used to create functional optical waveguides for illumination and sensing applications, such as touch sensing interactive displays using surface-printed light guides (left). Republished with permission of ACM (Association for Computing Machinery), from [54]; permission conveyed through Copyright Clearance Center, Inc. Freeform waveguides with optimized tapers and coupling (right) are possible at larger print areas compared with 2PP. Reprinted with permission from [56] © The Optical Society. (c) Printed FDM fiber optic preforms show the potential to extrude meters of fiber with complex cross-sectional geometries, down to the scale of optical wavelengths. (c) (left) The drawn structured optical fiber end-face illuminated at 630 nm. Reproduced with permission from [42] and SPIE. (c) -–(right) Extruded multimode fiber (L = 65 m) spooled and illuminated at 543 nm. Reprinted with permission from [74] © The Optical Society. (d) Free-standing, biocompatible, and curved waveguides made from silk fibroin were directly printed with DIW by [75]; Copyright Wiley-VCH GmbH, Weinheim. Reproduced with permission. Single mode illumination of curved waveguides (left) and the resulting fiber end face (right) at 633 nm are shown. (e) High-resolution 2PP is compatible with the integrated circuit fabrication, enabling manufacture of all-optical and hybrid electro-optical circuits. An early demonstration of printing optical circuits printed directly on an IC features waveguides elevated by printed supports [76], reprinted with permission.

Download Full Size | PDF

1. Direct Ink Writing

Direct ink writing has the capability to produce optics with detail and roughness on the scale of microns and nanometers, respectively. The performance and repeatability of direct ink writing depends on multiple variables, including material density and dispensing height, pressure, speed, and nozzle diameter. In particular, Dingeldein et al. optimized dispense height for different mixtures of photopolymers, observing that small variances in vertical position of the nozzle (e.g., 10 µm) could result in irregular waveguide structure. By using different concentrations of polysiloxane inks, dispense pressure, and height, multimode waveguides with 50 µm and ${\sim}{{20}}\;{\rm{\unicode{x00B5}{\rm m}}}$ core diameters were fabricated with high aspect ratios and losses similar to standard lithography [29]. The high aspect ratio of the printed waveguides is crucial for efficient light coupling, especially with standard fiber connectors. With additional studies on repeatability and the optimization of variables, direct ink writing can provide a viable method for manufacturing optical-quality waveguides. Relative to conventional lithography, DIW is resource efficient, as it does not require masks and utilizes low polymer volume, while allowing fabrication across large substrates.

Single mode and multimode waveguides can be created with DIW. Parker et al. produced free-standing single mode waveguides at 633 nm with approximately 5 µm core diameter and nanometer-scale roughness by using silk fibroin ink core material and a custom glass nozzle fabricated with a pipette puller [75]. Slight deformation of the cross-section occurred due to surface wetting on the borosilicate glass substrate. Despite the non-circular cross-section, good light transmission was achieved with optical losses for curved waveguides (${\lt}{0.81}\;{\rm{dB/cm}}$). Most inks flow freely following deposition, making the use of photo-curable polymers, which have low optical loss in the visible and near-IR wavelengths, challenging. Fugitive inks, which are UV-transparent sacrificial support or guide materials (acting like a positive resist), have recently been paired with photo-polymers to create multimode fibers with DIW. Lorang et al. applied a copolymer fugitive ink to contain photocurable OrmoClear until curing by using custom concentric nozzles. The fugitive ink was then thermally removed. Although the waveguides were 100s of microns in diameter, the fugitive ink micelle dimension produced 10–20 nm surface roughness, resulting in loss and quality similar to that achieved via 2PP and SLA [52]. It was also shown using the cutback method that optical losses in waveguides with different radii of curvature could be minimized by curing the ink directly following printing and increasing UV exposure intensity [52]. Figure 3 shows the results, wherein DIW was used to create freeform geometries with minimal post-processing. Depending on the application, DIW print resolution on the order of tens of microns can provide a highly functional waveguide. For instance, Hajj-Hassan et al. printed the negative of an array of ${{100}}\;{\rm{\unicode{x00B5}{\rm m}}} \times {{100}}\;{\rm{\unicode{x00B5}{\rm m}}}$ waveguides using commercially available photopolymers and successfully applied the completed array for luminescence-based biochemical oxygen sensing at 470 nm with a subsequently printed doped sol–gel [51]. The array utilized guided evanescence wave excitation with sensitivity and operation over the full scale of 0–100% O₂ concentration. Importantly, the minimal steps involved in printing the waveguide sensing platform implies the possibility of cost-effective and scalable manufacturing. To reduce feature size and waveguide diameter further, sub-micron and micron-scale commercially available needle tips could be used [29]. With the print specifications of DIW at or slightly larger than typical optical waveguide diameters, demonstrated applications have primarily been in illumination and sensing. Advanced ink design, such as those discussed above [52,75], in combination with commercially available sub-micron and micron-scale nozzles, are needed to push the technology toward high-quality optical imaging.

 

Fig. 3. Direct ink writing of two material optical waveguides adapted from [52]. Copyright Wiley-VCH GmbH, Weinheim. Reproduced with permission. (a) Straight waveguides with identical geometries used to characterize the optical loss reported in (b) with different LED exposures. (c) Curved waveguides with different radii of curvature. (d) Optical loss for each curvature printed. (e) Waveguide network with six different waveguides and three illumination sources to highlight minimal crosstalk.

Download Full Size | PDF

2. Inkjet Printing

Functional single and multimode fibers at optical wavelengths have been manufactured with inkjet printing. Ideally, waveguides should have high refractive index cores, high aspect ratios, and low transmission losses. By varying the number of vertical layers and temperature, Klestova et al. created a single mode fiber for transmission at 1.55 µm (telecom C-band) with a refractive index of ${\sim}{1.8}$ using a polymer matrix of titanium dioxide nanoparticles as the core material [77]. While the cross-section was planar (${{100}}\;{\rm{\unicode{x00B5}{\rm m}}} \times {{250}}\;{\rm{nm}}$), mean transmission loss was 3.52 dB/cm, and a splitter designed with the formulation presented 1 dB loss per split. Samusjew et al. improved the aspect ratio of a visible wavelength multimode fiber (80 µm width) by tuning the substrate surface properties [55]. Specifically, under-baked PDMS was optimized as the substrate for a stretchable polyurethane acrylate ink, which was deposited vertically until an 86° contact angle was achieved. Stretchable and biocompatible waveguides are of interest for soft robotics and biophotonics such as on-body sensing. The resulting fibers could be removed from the substrate and showed only 0.7 dB attenuation for every 10% applied strain at visible wavelengths. Printing junctions with smooth features and low loss is challenging with inkjet printing. Theiler et al. demonstrated printing of multimode waveguides (${{120}}\;{\rm{\unicode{x00B5}{\rm m}}}\;{\rm{x}}\;{{31}}\;{\rm{\unicode{x00B5}{\rm m}}}$) using capillary forces for printing smooth taper and junction features [56]. Transmission losses of ${0.61}\;{{\pm}}\;{0.26}\;{\rm{dB/cm}}$ at 650 nm were measured with some samples approaching losses achieved with photolithography and etching.

Due to a resolution of tens of microns and ${\gt}{{10}}\;{\rm{cm}}$ build volume, inkjet printing is well suited for applications in illumination and interactive displays. Willis et al. used inkjet to print 3D light pipes with 250 µm core and 84 µm cladding embedded into interactive devices [53]. Printing on a variety of arbitrary surfaces, as well as printing air pockets with minimal shape distortion, was achieved. Although losses were high for curved fibers, this initial attempt showed promise. Pereira et al. printed similar multi-material 1 mm core light guides also using VeroClearTM as a core material [54]. Hexagonal cross-sections were shown to have no loss in transmission compared to circular cross-section waveguides, while providing improved tiling for displays. Support material was selected from multiple cladding materials for its high absorption and low crosstalk, exhibiting the limited material choice in using inkjet. Additional development of inks is needed to enhance refractive index contrast and light transmission.

Photonic crystal structures for low loss and photonic bandgap applications have also been demonstrated in the microwave and terahertz regimes, showing the possibility of printing complex optical waveguides in the future with resolution improvements. Pouya et al. printed a tunable gyroid photonic crystal in the microwave range with rubber-like TangoBlack^TM and showed a compression-dependent change in the separation of transmitted spectral bands [57]. Ma et al. demonstrated hollow-core photonic bandgap fibers with 15% bandgap and 0.1 dB/cm transmission loss in the terahertz regime [58]. Cylindrical dielectric reflectors were used to create a larger bandgap over typical periodic reflectors.

3. Two-Photon Printing

Due to the small curing voxel of 2PP systems, 2PP techniques surpass other 3D printing methods for the fabrication of single mode fibers and subwavelength structures. Low roughness and high aspect ratios can be achieved, as well as custom cross-sections. These properties lend 2PP, also called direct laser writing (DLW), to applications in light-based integrated circuits (ICs) and precise manipulation of the light field. 2PP enables high density printing and 3D arbitrary trajectories, while offering CMOS compatibility and the use of broadband transparent polymers, qualities required for IC integration. Schröder et al. showed initial feasibility for printing single mode polymer waveguides on the IC surface [76]. 50 µm waveguides with 1.4 µm diameters were optimized for transmission by adjusting the diameter and distance to the substrate at 1.55 µm and for crosstalk. Air cladding and spatial separation were demonstrated to prevent crosstalk in simulation, showing that a 4 µm separation between parallel and perpendicular waveguides yielded 1% and 0.02% transmission, respectively, at 1.55 µm. Alternatively, insulating coatings, such as ${\rm{SiO_2}}$, have been previously used to preserve refractive index contrast and prevent crosstalk.

Planar lithography cannot easily fabricate 3D elements, such as polarizers and freestanding tunable and optomechanical components. Schumann et al. combined 2D planar lithography and 3D 2PP to demonstrate waveguide bridges between 2D waveguides and their use as polarizing elements in IC applications [78]. Although multiple post-processing steps were required, waveguide bridges had good mechanical stability and a roughness of 5 nm was achieved with ${\lt}{{3}}\;{\rm{dB/cm}}$ loss over 100 nm bandwidth at 1550 nm. Polarizing waveguide bridges were printed with 75% rotation efficiency and a loss of 5–7 dB, as well as a microdisk resonator coupled by bridges. Landowski et al. investigated the coupling efficiency of 2PP waveguides using out-of-plane coupling for high density optical chips [79]. Rectangular facets oriented perpendicular and parallel to the substrate were printed on single mode waveguides with bend radii down to 40 µm for structural stability. Parallel facets showed higher coupling efficiency, and the configuration yielded 10 dB insertion loss and transmission loss ${\lt}{0.81}\;{\rm{dB/cm}}$ at 780 nm. Both stadium waveguides with lengths up to 3.8 mm and “${{y}}$”-splitters with adjustable ratios were demonstrated. Nguyen et al. demonstrated single mode embedded waveguides from a single custom material [80]. A refractive index contrast of 0.013 was achieved through multiple curing steps with intermediate diffusion of a gaseous monomer into the cladding. The resulting waveguides could be printed in high density and in multilayers. A transmission loss of 0.37 dB/cm was measured at 850 nm with a numerical aperture (NA) of 0.148 and symmetric mode fields. In a follow-up study [81], a three-waveguide fan-out was demonstrated with 34 dB crosstalk between waveguides. The refractive index profile was further studied and the addition of a photosensitizer to the polymer was shown to reduce roughness by a factor of seven. This work shows potential for printing of fully optical ICs but limits incorporation of electrical elements. On-chip 2PP printing also finds use in low resource sensing applications. Cadarso et al. showed this potential [82], designing a “nanofence” subwavelength waveguide array for the evanescent field sensing of lead. High aspect ratio polymer waveguides (${{250}}\;{\rm{nm}}\;{{\times}}\;{{6}}\;{\rm{\unicode{x00B5}{\rm m}}}\;{\times}\;{{1}}\;{\rm{mm}}$) were printed in an array to create a cumulative, multimode, extended evanescent field, which allowed 90% of guided light to interact with the medium. The array measured 12 dB/mm losses at 635 nm for a 20 nanofence array and a detection limit of  7.3  nM in lead sensing.

2PP can facilitate precise manipulation of signal phase and amplitude. As such, it has been used to design waveguide arrays for phase-shifting and spectrally selective applications. Belle et al. demonstrated fabrication of hollow waveguide arrays that act as quarter wave plates [83]. The negative of the rectangular ${{4}}\;{\rm{mm}}^2$ waveguide array was printed with a height of 2 µm, and the resulting array was coated with gold to induce the phase shift. The resulting phase shift differed from the target value due to rounded edges but achieved similar results to standard technologies with 61% transmission at 1550 nm. By changing the array height, the phase shift could be adjusted. Space variant polarization control was also simulated. To produce a large array, Belle et al. utilized stitching of adjacent writing fields to overcome the generally small writing area available from 2PP. Due to this limitation and limitations of printing time, 2PP may need to be combined with other printing methods for macroscale systems requiring high optical quality.

Large diffraction angles are needed for beam steering and holographic applications. Zheng et al. printed a static optical phased array for holography with widths from 300–700 nm to enable a pixel size close to visible wavelengths [84]. Using ${1.5}\lambda$ spacing between waveguides to balance coupling and angle, a relatively large diffraction angle of ${\sim}{{40}}^\circ$ was achieved, showing potential for pixel-wise manipulation. Bragg grating waveguides have also been fabricated with 2PP. Spectral width and selection are dependent on the refractive index contrast, the homogeneity of spatial period, and the quality of interfaces and mode coupling. Goraus et al. printed a Bragg grating waveguide with 70 µm support pillars to enable direct fiber coupling for on-chip applications [85]. IP-Dip polymer was used for its mechanical properties but resulted in a slight shift of the reflectance wavelength to 1543 nm from 1550 nm due to shrinkage. The waveguide was layered with a spatial period of 1.59 µm, a cross-section of ${{2}}\;{\rm{\unicode{x00B5}{\rm m}}} \times {{10}}\;{\rm{\unicode{x00B5}{\rm m}}}$, and a transmission dip of 15 dBm. Phase-shifted Bragg gratings have also been created for sensitive ultrasound detection by Wei et al. [86]. Printed waveguides achieved a 50% duty cycle, a 20 nm stopband, a 0.085 nm full linewidth and minimum transmission loss of 15 dB with dimensions of ${1.5}\;{\rm{\unicode{x00B5}{\rm m}}} \times {{2}}\;{\rm{\unicode{x00B5}{\rm m}}} \times {{100}}\;{\rm{\unicode{x00B5}{\rm m}}}$ and a grating depth of 250 nm. The stopband was reduced from the design due to rounded edges from curing. The resulting Q factor of ${\sim}{{18}}\;{\rm{k}}$ was low but demonstrated sensitive detection. In a similar manner, Li et al. fabricated 1D photonic crystals (${{48}} \times {{48}} \times {{30}}\;{{\unicode{x00B5}}}{{\rm{m}}^3}$) at 1550 nm using alternating compact and 24% fill-fraction layers with the same monomer [87]. The design showed 90% reflectance in the near IR, demonstrating potential as a specular reflector, and was tunable by choice of fill-fraction and layer thickness. Imperfections in printed layer thickness uniformity were noted to cause asymmetric shift in the bandgap and document a needed improvement in consistency of 2PP.

Two-photon printing techniques are capable of creating structures of subwavelength dimensions in the visible range, creating new opportunities for precision optical illumination and light-guiding. Palima et al. took 2PP a step further in light manipulation by demonstrating selective illumination and focusing beyond the diffraction limit [88]. Using optical traps, 1.5 µm diameter 2PP-printed waveguides with 0.4 µm tapers were oriented to couple light from a low-NA objective, redirect, and selectively excite fluorescence in single microbeads from 1D and 2D groups of microbeads. Shi et al. explored the potential of 2PP for quantum optics ICs, performing localization of nitrogen valences, followed by selective 3D lithography aligned to the single photon emitters with a custom optical system and custom, solid photoresist [89]. The photoresist and crossed arc waveguides were designed to exhibit low fluorescence background while optimizing collection efficiency. Specifically, excitation light was guided orthogonally relative to emitted photons, enabling a measured collection efficiency of ${\sim}{{30}}\%$ from the excited single photon emitter after arc radii optimization. Thus, small-scale optics made by 2PP may help bridge the micro-to-nano gap to measure nanoscale targets in the future. 2PP-printed waveguide components are shown in Fig. 4.

 

Fig. 4. (a) Nanofence structure. Reprinted with permission from [82]. Copyright 2016 American Chemical Society. (b) Waveguides interfacing with microdisk resonator from [78], reprinted under a Creative Commons 3.0 license,http://creativecommons.org/licenses/by-nc-sa/3.0/. (c) Y-splitting waveguide from [79], reprinted under a Creative Commons 4.0 license, http://creativecommons.org/licenses/by/4.0/. (d) Schematic of waveguide reorganization, (e) input of the waveguide, and (f) output of the waveguide after reorganization. Reprinted from [81] with permission from Elsevier. (g) Bragg waveguide on supports. Reprinted from [85] with permission from Elsevier.

Download Full Size | PDF

4. Fused Deposition Modeling Printing

With typical resolutions ${\gt}{{100}}\;{\rm{\unicode{x00B5}{\rm m}}}$, it is difficult to produce high performance optical waveguides by directly using FDM-based 3D printing. One direct application of FDM is for printing preforms that can be used with traditional fiber drawing methods to achieve microscale diameters. A multimode, step-index fiber was drawn from a printed preform by Cook et al. [74], achieving 0.44 dB/cm transmission loss at 1546 nm. Slow printing speeds and high infill percentages were used to maximize quality of the preform and minimize air bubbles during the drawing process. Horizontal printing of the preforms resulted in asymmetry in the drawn fibers, a trade-off allowing the use of the higher $z$ resolution. Vertical printing can be used to create greater cross-sectional symmetry but results in reduced cross-sectional resolution. Zhao et al. demonstrated that 3D printed preforms could be used to make fibers with arbitrary cross-sectional geometry [43]. Complementary cladding preforms were printed to create “special shape” fibers with circular, square, triangular, and rectangular cores. Core and cladding were drawn together in a low temperature polymer draw tower, producing core dimensions on the order of 40 µm and down to 8.5 µm. Optical transparency of the preforms was found to depend on print temperature, layer height, and print speed. At 633 nm, the fibers exhibited losses from 0.7–1.8 dB/cm. Both traditional drawing towers and 3D printers have been used as micro-draw towers for 3D-printed preforms.

Canning et al. characterized the nozzle of several low-cost 3D printers, showing that the temperature distributions were consistent and allowed direct fiber drawing using the filament as a preform [41]. Using the 3D printer as a microfurnace, single material, air core fibers with diameters of 30–40 µm were fabricated and achieved a transmission loss of 0.26 dB/cm (approaching standard optical fiber losses) at 543 nm and 1550 nm for polyethylene terephthalate glycol (PETG) fibers. Talataisong et al. used a custom nozzle to directly draw a micro-structured polymer optical fiber with an FDM printer, showing that the microstructure could be maintained after drawing [90]. A functional fiber was attained with 1.1 dB/cm transmission loss at 1557 nm. Meters of fibers can be drawn depending on the drawing tower used. Another way FDM may disrupt conventional fabrication methods is by shortening the “stack and draw” process used for making faceplates. Wang et al. printed 20k, 280-µm-diameter polymer fibers for direct image transfer, directly using FDM and attaining 1.78 LP/mm resolution [9]. With air as the cladding, reduction of crosstalk was explored by printing alternating perpendicular fiber layers, but there was a trade-off with transmission and spatial resolution. Thus, the printed faceplate was concluded to be effective for light transfers of several centimeters.

Structured optical fibers can also be fabricated through the drawing process by printing their complex geometries on the macroscale as a preform with FDM. Cook et al. printed a polystyrene preform with a single ring of air holes, demonstrating the possibility [42]. Annealing of the preform improved its transparency. Drawn fiber transmission was ${\gt}{{30}}\%$ with measured losses of 0.75–1.5 dB/cm at 632 nm, 1064 nm, and 1550 nm. However, the resulting cross-sections were asymmetric, and inaccuracies due to low resolution were noted to compromise light confinement in the core. While polymer fibers are gaining popularity as biocompatible optical materials, the materials result in lossy waveguides in the terahertz regime. Propagation loss may be minimized by using air core fibers. Using hollow core fibers for terahertz guidance, 95% of modal energy can be confined [91]. In the terahertz regime, hollow core bandgap, Bragg, and antiresonant fibers have been demonstrated and discussed in detail by Cruz et al. [91]. Antiresonant fibers printed directly by FDM have shown losses as low as 0.1 dB/cm. In particular, Cruz et al. achieved broad mode guidance from 0.1–1.1 THz from a printed negative curvature antiresonant fiber. FDM offers greater flexibility and convenience for hollow core fiber fabrication compared with conventional fabrication methods [91].

B. Lenses and Lens Arrays

Individual lenses, compound lenses, objectives, and lens arrays are the most fundamental and functional components of virtually any detection or imaging application. Depending on their function, they need to provide different levels of form correction (coupling on-axis lenses and illumination optics will have lower requirements compared to imaging optics). For example, a diffraction-limited imaging system needs to provide ${0.07}\lambda$ root-mean-square (RMS) wavefront ${\rm{error}}/{0.25}\;\lambda$ peak-to-valley error (or alternatively ${\gt}{0.8}$ Strehl Ratio) over a designated FOV. Surface quality needs to be at nanometer level roughness (10 nm for molded lenses is common [92]), while the density of the material has to be highly uniform and stress-free to avoid both wavefront error and birefringence effects. This places high requirements on the selected fabrication process, and thus also determines possible applications for the technique. For example, FDM methods struggle with form, roughness, and material isotropy, making FDM best suited for longer wavelengths/THz applications. FDM, however, allows large printing volumes at short production times. On the other end of the spectrum, 2PP methods produce miniature components at low roughness and wavefront error permitting designs with diffraction-limited performance. It should be noted that the focus of many recent publications is to demonstrate the ability to print new types of components, and thus they are not fully optimized or characterized. Additional effort in metrology is necessary to both provide a complete characterization of the printed components and provide feedback for the fabrication process.

Additive technologies provide an attractive method of lens and lens array fabrication. 3D printing allows for freeform design of lens surfaces, development of compound optical systems with no need for future alignment, and integration with other components such as printing directly on the IC surface, integration with fiber couplers, and kinematic alignment features. Below, we briefly review different 3D printing methods in the context of their imaging/light collection applications and operational parameters. As a reference, Fig. 5 demonstrates the detail level and surface roughness obtained with different 3D printing technologies.

 

Fig. 5. Comparison of the 3D printing methods in the context of achievable detail size and surface roughness. The component images from left to right. FDM: printed fiber faceplate, reprinted with permission from [9] © The Optical Society; GRIN lens, reprinted from [36] underr a Creative Commons 3.0 license, https://creativecommons.org/licenses/by/3.0/; and step-index optical fiber drawn from 3D printed preforms, reprinted with permission from [74] © The Optical Society. SLA: printed lens, reprinted with permission from [64] © The Optical Society; solar concentrator array, reprinted from [93] under a Creative Commons 4.0 license, http://creativecommons.org/licenses/by/4.0/; 12.7 mm diameter, ${-}{{25}}\;{\rm{mm}}$ focal length plano-concave lens fabricated using Form2 SLA printer. Inkjet: lens array, reproduced from [94] with permission. Copyright SPIE; printed lens [95], reprinted with permission; printed waveguide reprinted with permission from [55]. Copyright 2017 American Chemical Society. 2PP: compound microlens system reprinted from [22]. © The Authors, some rights reserved; exclusive licensee AAAS. Distributed under a CC BY-NC 4.0 license http://creativecommons.org/licenses/by-nc/4.0/; SERS body [96] Copyright Wiley-VCH GmbH, Weinheim. Reproduced with permission; fiber-to-chip coupler [97], reprinted under a Creative Commons 4.0 license, http://creativecommons.org/licenses/by/4.0/; and achromatic axicon, reprinted with permission from [12] © The Optical Society.

Download Full Size | PDF

1. FDM Printing

Lenses produced using FDM technology generally work in the THz regime due to nozzle resolution limits. Graded refractive index (GRIN) and Fresnel lenses have been particularly prevalent. Hernandez-Serrano et al. showed a GRIN lens design using polystyrene and air to control the refractive index gradient [39]. A similar concept was shown by Acikgoz et al. who used gradient relative permittivity to create a GRIN lens. This concept allows for the construction of a GRIN lens using a single material by introducing holes in the material. By changing the holes-to-material ratio, the refractive index of the material can be changed [38]. Likewise, a few efforts have been made to fabricate diffractive lenses using FDM printing [37,40]. Selected examples of lenses fabricated using FDM technologies are shown below in Fig. 6.

 

Fig. 6. THz regime components produced by FDM. (a) GRIN lens fabricated by Zhang et al. [36]. (b) Fresnel lens fabricated by Zhang et al. [37]. Each image is reprinted under a Creative Commons 3.0 license: https://creativecommons.org/licenses/by/3.0/.

Download Full Size | PDF

2. Inkjet Printing

Similar to FDM, inkjet printing has largely been limited to the THz regime [34,35] due to print resolution. Lenses for optical wavelengths produced using inkjet printing typically suffer from roughness or form errors. Sung et al. demonstrated a simple method for mass fabrication. However, lens quality is highly dependent on environmental factors, such as the temperature of the print surface [59]. Zhu et al. utilized a different approach, combining inkjet printing and imprint lithography to fabricate flexible biconvex microlens arrays [61]. For a detailed accounting of microlens fabrication with inkjet printers, see Alaman et al. [60]. As described in the technology description in Section 2, macroscale inkjet-printed optical components have emerged on the market for ophthalmic applications [62]. These components reportedly can have a surface roughness as low as 10 nm, while producing components that are 10–15 mm in diameter [94]. Quality of the surface depends on a number of features, including the radius of curvature of the surface, the size of the component, and the contact angle between the printed component and the substrate it prints onto [17]. In direct comparison to commercial glass lenses of matching specification, inkjet-printed lenses displayed roughness average (Ra) values below 20 nm, which is comparable to commercial molded optics. However, RMS wavefront errors for printed lenses were significantly larger than the errors in the glass equivalents. To address the physical resolution limitations presented by a single print, an iterative method for improving the quality was developed by Assefa et al. [7]. The iterative process relies on measuring the wavefront error after printing from a specific design. The error is quantified, and the surface of the design is adjusted to account for material shrinkage during polymerization and overall wavefront errors. After measurement, the process is repeated. With 5–10 iterations of this process, wavefront error values were reduced from several wavelengths to sub-wavelength range. These results show that while inkjet technology can potentially fabricate lenses with comparable imaging performance to commercial lenses, it requires an iterative process to correct for form errors.

3. Vat Polymerization–SLA and DLP

SLA printers can also be used to fabricate singlet optical lenses of few to tens of millimeter diameters [64,98]. These lenses are finished using post-processing methods that incorporate additional printing resin to smooth out lens surfaces after the printing process. In one application of this technique, plano-convex lenses are post-processed by spin-coating printing resin onto the exterior surfaces of printed lenses. Using this post-processing technique, average RMS roughness values between 13 and 28 nm could be achieved with average RMS wavefront error values between ${0.30}\lambda$ and ${0.37}\lambda$. A secondary post-processing technique involved using negative matching glass commercial components as a cheap mold to fabricate small batches of printed lenses. Using this method, an average RMS roughness value of 6 nm and a RMS wavefront error of ${0.05}\lambda$ were achieved [64]. SLA-printed components can also allow for the incorporation of alignment features and easy assembly of multiple parts. Additional efforts have focused on adapting projection micro-stereolithography for optical fabrication. By using meniscus equilibrium post-curing and grayscale photopolymerization, lenses up to 3 mm in diameter with roughness values under 10 nm could be produced, as shown by Chen et al. [99]. This same group further iterated this process, replacing the resin recoating step with a continuous process and substituting Teflon membranes for polydimethylsiloxane (PDMS), resulting in a process that could produce 3 mm tall lenses in 2 min. This resulted in aspheric lenses with an imaging resolution of 3.1 µm [100]. Digital light processing (DLP) has also been investigated for the production of freeform refractive elements. By creating DLP-produced molds, PDMS can be cast to the mold and peeled off to create a finalized structure to create custom designed light patterns [101].

4. Two-Photon Printing

The high resolution available with 2PP can lead to a wide variety of printable lens designs. Kiekens et al. demonstrated the ability to fabricate a single plano-convex lens on a glass substrate for use in endoscopic applications. The fabricated lens displayed an RMS total form error of 595.5 nm over the 500 µm diameter [102]. Other groups have shown how to print lenses onto a variety of application-driven substrates. Thiele et al. demonstrated the printing of collimation elements directly onto an LED chip, reducing the size of collimation optics typically seen on LED chips and allowing for the creation of custom irradiance patterns [103]. Microlens arrays can be fabricated directly onto multi-core fiber tips for improved coupling efficiency and fill factor, as shown by Dietrich et al. [104]. By etching marking elements into the substrate, Sartison et al. were able to create a solid immersion hemispherical lens to cover a quantum dot, which improve signal-to-noise ratio and increased localization performance [105]. Immersion lenses for quantum dots were also investigated by Bogucki et al. By printing a large number of lenses in a rectangular grid, it was possible to find different emitters and improve signal [106]. Rather than use an immersion lens, Johlin et al. used algorithms to design a freeform nanophotonic lens that improves the directivity of emitting nanowires [107]. It is also feasible to print diffractive lens structures using two-photon printing. Li et al. show a new design of a diffractive cylindrical lens in which the thickness of the lens decreases with increasing radial distance [108]. Finally, Shäffner et al. used the Nanoscribe system to create microlens arrays that were selectively turned on and off to create controllable optical tweezers with the array able to be designed to match the needs of the application [109]. Some examples of these components are shown in Fig. 7.

 

Fig. 7. Example of components printed using two-photon printing. (a) LED collimation optic on an LED chip, reprinted with permission from [103] © The Optical Society. (b) The solid immersion lens printed over etched alignment elements from [105]. (c) Grid of immersion lenses from [106]. (d) Left, Nanophotonic lens from [107] and right, computer-generated design of the same lens. Scale bar 2 µm. (e) Diffractive cylindrical lens surface from [108]. Each image sourced from [105]–[108] is reprinted under a Creative Commons 4.0 license, http://creativecommons.org/licenses/by/4.0/.

Download Full Size | PDF

One of the major benefits to the 3D printing of optics is the ability to print components that will be in alignment after the printing process. Several studies have been published where multiple components have been fabricated using two-photon printing. Schmid et al. showed the ability to create achromatic elements composed of multiple photoresists using the GT2 system. Both an axicon and a doublet were produced that corrected for chromatic effects by printing the first portion of the print in one resist, developing the print, and then printing the second portion in the new resist [12]. The same group also demonstrated the ability to create diffractive lenses printed in alignment, with the benefit of much quicker printing times due to the relative thinness of diffractive lenses in comparison to refractive lenses [110]. The feasibility of producing diffractive lenses in alignment on the ends of fibers was also proven, as this setup was used to produce optical tweezers [111]. Furthermore, the ability to create spherical and toric lenses, as well as freeform optics and polarization control elements, directly on the surface of single mode fibers has also been demonstrated [11]. The use of two-photon printing to create these components gives the ability to design multi-component optics with each component separated by air. The functionality of producing multiple components in alignment allows for the creation of fully printed multi-element systems. Thiele et al. demonstrated this by creating multiple examples of doublet objectives, which were paired with a CMOS sensor to create a foveated imaging camera [22]. Individual components produced using 2PP measured using atomic force microscopy displayed an RMS roughness value of 43 nm [11].

Multi-element systems were further explored by printing a micro-objective to work in tandem with a quantum dot and microlens (fabricated via electron beam lithography) to improve photon-extraction efficiency in order to create an efficient way to transfer light from a single quantum dot to external optical components [112]. Another example is provided by Tadayon et al. who demonstrated the design of probes for deep tissue imaging. As an alternative to printing in two different materials to get a refractive index change in the system, lens structures are printed in one resist with designed cavities that can be filled after printing by a material with desired refractive index properties [113]. Li et al. further show the ability to print components in alignment by demonstrating a methodology to print microscale alignment features such as fiber guides, which can work with nanoscale printed lenses as a mechanism to ensure alignment [114]. A potential downside of printing an entire system using a two-photon printing setup is that the entire system, including supports, will be made of transparent material, eliminating the possibility of directly printing apertures into the system and allowing for the possibility of stray light. To address this problem, Toulouse et al. demonstrated a scheme in which microfluidic cavities were printed into the system supports, which could be filled with non-transparent ink, creating apertures in the system [15]. Microlenses printed using two-photon lithography can also be used in collaboration with mirror structures, which can be fabricated through a combination of printing and metal coating. Cao et al. show the usefulness of a microlens-mirror system and microfluidic channels for single droplet analysis [115]. Finally, He et al. were able to use a Nanoscribe system to develop an adaptable liquid crystal microlens array by first printing a microlens array and then adding a liquid crystal solution to the cell [116]. Multi-element components fabricated using two-photon printing are shown in Fig. 8.

 

Fig. 8. (a) Achromatic axicon and (b) achromatic doublet, reprinted with permission from [12] © The Optical Society. (c) Triplet stack of diffractive lenses, reprinted with permission from [110] © The Optical Society. (d) Doublet diffractive lenses on fiber tip for optical trapping; scale bar = 100 µm. Reprinted with permission from [111]. Copyright 2020 American Chemical Society. (e) Triplet on fiber tip from [11], reprinted under a Creative Commons 4.0 license, http://creativecommons.org/licenses/by/4.0/. (f) Doublet micro-objective for foveated imaging (Reprinted from [22]. © The Authors, some rights reserved; exclusive licensee AAAS. Distributed under a CC BY-NC 4.0 license http://creativecommons.org/licenses/by-nc/4.0/”).

Download Full Size | PDF

C. Fabrication of Grating/Diffractive Structures

Most efforts to create diffractive structures utilize 2PP techniques because they can provide the necessary detail level. Several efforts to print grating structures have been recently published. Purtov et al. report on the fabrication of grating structures composed of nanopillars. By varying the laser power used for polymerization, pillars with diameters down to 184 nm could be achieved [117]. Cheng et al. demonstrated the creation of subwavelength gratings with periods of 314 nm to focus reflections in the near field using two-photon printing [118]. Instead of focusing in the near field, Ni et al. produced a super-oscillatory lens to focus light through diffraction in the far-field region [119]. Mu et al. also pursued focusing with grating structures, creating metallic pyramidal probes by using metal deposition post-printing that incorporated asymmetrical gratings of 400 nm periodicity and finding that the grating structures greatly improved focusing [120]. Two-photon printing allows for the fabrication of complex shapes that incorporate grating structures. Hsiung et al. utilized this capability to design biomimetic grating structures with a 670 nm period based on spider scales to efficiently separate light into different wavelengths with superior performance when compared to a traditional 2D grating structure [121]. Xiao et al. demonstrated a ${{3}}\;{\rm{mm}} \times {2.4}\;{\rm{mm}}$ phase mask that could effectively split sunlight into an infrared and visible band for use in photovoltaic cells [122]. Grating structures fabricated using 2PP have also been used to introduce color filtering into systems due to color-dependent diffraction into higher orders. By varying the height and spacing of the pillars in a rectangular grid, Nawrot et al. were able to tailor grating structures to act as color filters [123]. Some grating-based structures are shown below in Fig. 9.

 

Fig. 9. (a) Biomimetic grating structures based on spider scales; scale bar = 2 µm. Reprinted from [121] under a Creative Commons license, https://creativecommons.org/licenses/by/4.0/. (b) Pyramidal probes with asymmetrical gratings reprinted with permission from [120] © The Optical Society. (c) Phase mask structure. Reprinted with permission from [122]. Copyright 2016 American Chemical Society. (d) Subwavelength grating for focusing in near field, reprinted with permission from [118] © The Optical Society.

Download Full Size | PDF

D. Fabrication of Phase Modulation Elements

Phase modulation elements are another prominent use of two-photon printing due to the speed, ease, and shape complexity available to fabricate them [124]. Gissibl et al. printed phase masks directly onto fiber tips to achieve desired spatial intensity patterns [125]. Rather than use fiber tips, Lightman et al. printed phase elements on the planar side of aspheric lenses [126] and on nonlinear crystals [127] to generate desired intensity distributions. Lastly, the fabrication of components to create angular orbital momentum beams has also been demonstrated. Spiral phase plates have been printed directly on a fiber tip [128] and on the surface of a MEMS filter [129]. Additional work has focused on metamaterial fabrication. Hahn et al. developed a multi-focus two-photon printing system to create 3D chiral metamaterials [130]. Lio et al. incorporated metamaterials directly into the printing process to improve resolution. This allowed for the fabrication of ultrathin (30 nm in height) dielectric metalenses [131]. Two-photon printed phase elements are shown in Fig. 10.

 

Fig. 10. (a) Phase element on fiber tip from [125]. (b) Zoomed-in image of same phase element. (c) Phase element on nonlinear crystal from [127]. (d) Phase elements on aspheric lens from [126], scale bar = 25 µm. Each image is reprinted with permission, © The Optical Society.

Download Full Size | PDF

E. Other Photonic/Optical Applications

Multiple efforts have been undertaken to incorporate 2PP into the production of surface-enhanced Raman spectroscopy (SERS) sensors. Braun et al. created a SERS infrared absorption sensor by printing a mold for nano-antennas. After printing, the substrate is coated with metal. The photoresist is then removed leaving only the metal deposits, which can be used to perform various sensing applications [14]. Kim et al. used 2PP and subsequent metal deposition to create a SERS probe on a fiber tip. This probe was then demonstrated to be functional for the detection of E. coli samples [132]. Similarly, Xie et al. also fabricated a SERS sensor on a fiber tip, fabricating both a dome-like mirror structure and a spherical SERS body by printing with a 2PP printer and then coating it with gold [96]. Other groups have examined photonic crystal production. Hu et al. examined the use of a custom photoresist, which allowed for switchable refractive index properties, through the fabrication of a photonic woodpile structure [13]. As with grating structures, Hsiung et al. again use biomimetic designs to fabricate non-iridescent photonic devices based on tarantula hairs as an alternative to synthetic dyes and pigments [133]. Lastly, designs have been proposed that incorporate two-photon printing with liquid crystals. Creating partition layers, support structures, and filling cavities with orthogonally oriented liquid crystals has allowed for the production of fast-response phase modulation elements [134,135]. Nocetini et al. used two-photon printing to structure liquid crystalline networks, resulting in structures that have controllable responses to light [136]. Additionally, filling a woodpile structure with liquid crystals has allowed for the creation of elements that can be used for spatial filtering and polarization [137]. A sampling of the components described in this section are shown in Fig. 11.

 

Fig. 11. (a) Cross spike array on fiber tip for SERS applications. Reprinted from [132] under a Creative Commons 4.0 license, https://creativecommons.org/licenses/by/4.0/. (b) SERS body from [96]. Copyright Wiley-VCH GmbH, Weinheim. Reproduced with permission. (c) Partition layer design, reprinted with permission from [134] © The Optical Society. Arrows indicate the orthogonality of the structure.

Download Full Size | PDF

4. COMMERCIALLY AVAILABLE SYSTEMS AND THEIR SPECIFICATIONS

While other portions of the paper have described some research level updates on various additive manufacturing systems, this section aims to describe the current state of printers that can be used in an off-the-shelf capacity. 3D printing advancement started in research labs through early feasibility demonstrations (custom 2PP systems) and through the adoption of existing 3D printing techniques in photonic applications (FDM and inkjet printers). The current state of technologies provides many available commercial 3D printing tools to be further merged and advanced. This section provides an abbreviated summary of functional parameters for selected (and representative) commercially available systems. Figures 5 and 12 provide comparison of different functional printer parameters. Figure 5 shows the achievable roughness and the smallest detail size obtained with different printing methods, while Fig. 12 presents characteristic print volume and printing time.

 

Fig. 12. Comparison of the print volume and print time for selected commercial 3D printers, representing four technologies: FDM, SLA, Inkjet, and 2PP. In the top portion of the graph, rectangular prisms show (to scale) maximum volumes for Ultimaker S5, form 2, Objet 24, and Nanoscribe GT2. The maximum potential volume for GT2 is ${{100}}\;{\rm{mm}}\;{\times}\;{{100}}\;{\rm{mm}}\;{\times}\;{{8}}\;{\rm{mm}}$, as enabled by stage range and as marked in the figure. In practice, this volume is difficult to achieve due to challenges of delivery in printing material and objective material immersion for tall structures. The printed components inside the volume prisms are at their approximate size to better represent scale differences. The bottom portion of the figure compares printing time of ${{100}}\;{\rm{mm}}\;{\times}{{100}}\;{\rm{mm}}\;{\times}\;{{8}}\;{\rm{mm}}$ volume using different printers. The printing time ranges from 2 h to several weeks. FDM image from [36], reprinted under a Creative Commons 3.0 license, http://creativecommons.org/licenses/by/3.0/; SLA image reprinted with permission from [64] © The Optical Society; 2PP image reprinted with permission from [12] © The Optical Society; and Inkjet image reproduced from [94] with permission. Copyright SPIE. The table below the image lists details on the printing settings used to generate estimates.

Download Full Size | PDF

A. FDM Printing

Several different commercial options exist for FDM printing. In general, this class of printing offers the least expensive commercial printers, with costs generally between 1,000 and 6,000 USD. Resolution is limited by the diameter of the printing nozzles, with nozzle diameters typically between 250 and 500 µm (for Ultimaker, Lulzbot, and Makerbot). As shown above in Fig. 12, this class of printers is also able to print larger components. As an example, the Ultimaker S5 has an available build area of ${{330}}\;{\rm{mm}}\;{\times}\;{{240}}\;{\rm{mm}}\;{\times}\;{{300}}\;{\rm{mm}}$.

B. Inkjet Printing

Stratasys offers a commercial line of inkjet printers under the commercial name Polyjet, although other commercial options are also available. Resolutions for commercial inkjet systems exceed those of FDM printers. For instance, the J850 Pro from Stratasys can achieve 14 µm resolution in certain printing modes [138]. This higher resolution comes at a correspondingly higher cost, with some commercial options exceeding 100,000 USD. Printing volumes for inkjet printers can handle fairly large prints, with the Objet24 having a build volume of ${{234}}\;{\rm{mm}} \times {{192}}\;{\rm{mm}} \,\times 148.{{6}}\;{\rm{mm}}$ [139].

C. SLA Printing

Commercial SLA systems have resolutions limited by the size of the laser spot they can achieve but generally achieve resolutions in between that of inkjet and FDM printing. In the vertical direction, resolutions of 25 µm are typical. Costs for these machines are on the order of thousands of dollars, with the Formlabs line of printers generally between 2,500 and 3,500 USD. Printing volume for the printers is feasible for most optical designs, with the Form 3 having a build volume of ${{145}}\;{\rm{mm}} \times {{145}}\;{\rm{mm}} \times {{175}}\;{\rm{mm}}$. SLA methods based on liquid crystal display (LCD) or DLP have lower resolution (100+ microns) but also enable shorter printing times.

D. Two-Photon Printing

A few different commercial 2PP systems are available on the market, with the most prevalent being the Nanoscribe GT2 and recent additions of the Quantum X and LithoProf3D-GSII from Nanoscribe and Multiphoton Optics, respectively. The GT2 system offers a maximum build volume of ${{100}}\;{\rm{mm}} \times {{100}}\;{\rm{mm}} \,\times {{8}}\;{\rm{mm}}$, with a minimum feature size of 160 nm. However, the time scale to print at full build volume generally limits this technology to the microscale, especially since the system’s objective needs to be continuously immersed in printing polymer. Quantum X and LithoProf3D-GSII offer prints over a larger area with continuous polymer dispensing. Industrial application of these printers includes freeform mold fabrication for volume component production. To increase speeds, the newer generation printers modulate the laser beam power. Costs for these printers are also the greatest, with systems often substantially exceeding 500,000 USD.

5. CONCLUSIONS AND DISCUSSION

As presented above, additive manufacturing methods are an emerging force in the area of optics and photonics. They allow freedom of design form, integration of approaches, and new design concepts.

There has also been a focus on 3D printing of glass-like material, which has not yet been discussed in this review. Here, we aim to provide a brief summary in the context of the future direction of this research. Within the technique of FDM, several efforts have been undertaken in this vein of research. Among these is the efforts by Zaki et al. who showed a method to print phosphate glass after pulling glass preforms through a draw tower [140]. This resulted in printed glass with layer thickness as low as 100 µm when using a modified FDM printer. Other work has focused on the printing of chalcogenide glass using FDM printing. Baudet et al. showed a method to create complex chalcogenide printed shapes using a customized printer [141]. This work was expanded on by Carcref et al. who demonstrated the printing of chalcogenide glass by also utilizing a modified FDM printer to create preforms for hollow core fibers [142].

Beyond FDM there is work to expand additive manufacturing for the production of glass structures. Luo et al. showed printing of fused quartz glass by heating glass filament with a laser, and their successful fabrication of 3D structures showed promise for future optical printing [143]. This same group further demonstrated this laser-heated filament process, demonstrating waveguides printed using single mode optical fibers as the filament in work published in Johnson et al. [144]. A similar printing methodology was investigated by Rettschlag et al. who examined the production of microspheres [145]. This same group also studied the fabrication of curved waveguides through this printing scheme, leading to the creation of functional waveguides [146]. A commercial company, Nobula, has been founded based on this printing method and has shown the ability to fabricate a variety of components [19].

The commercial company Glassomer has also been instrumental in glass printing research. This process works by using a photocurable silica nanocomposite that can be converted to fused silica glass via post-printing heat treatment [147]. This technology has found a home in both stereolithographic printing and two-photon printing. Chu et al. were also able to demonstrate silica optical fibers from a DLP printed preform. The cladding material was printed, and the core material was poured in after printing and then cured. Following this cure, the complex was put through a thermal debinding process, leaving behind the desired preform [148]. In the realm of two-photon fabrication, Wen et al. were able to show a method of fabricating silica in an effort to advance towards microphotonic silica structures. Printed silica resolutions were better than 200 nm [149].

This review also points toward opportunities and challenges. Challenges include expanding new materials and enabling printing with multiple materials. Printing in multiple materials is available when using an FDM printer with dual extruders. However, other methods struggle with such implementations and often need iterative or sequential processes. Further innovation to allow for dual material printing in commercial systems could further innovate optical fabrication. Dual material printing using a 2PP printer could enable easy doublet fabrication without requiring the printing of alignment markers and the multi-step development of parts. For example, using microfluidic structures in combination with a commercial system has allowed for multi-material printing when using two-photon printing [21]. Enabling a larger variety of materials will also enable increased functionality. New material development for all forms of printing could allow for a larger range of options for design. As an illustration, a larger library of available printing materials could lead to significantly more material Abbe number options when designing wavelength corrected parts.

As discussed in this review, all methods of printing have significant advantages but come with their own disadvantages. Two-photon printing brings the ability to print completed systems rather than single components, a limitation of other printing methods. However, 2PP presents important challenges when considering component printing times that practically limits print size. In contrast, FDM, SLA, and inkjet printing allow lower cost printing and larger volume production, with the drawback of lower printing resolution. Future developments in the field of additive manufacturing could see the development of hybrid systems, which find a way to balance the pros and cons of various printing technologies. For instance, SLA printing could be combined with 2PP printing in a single system to enable a system that is capable of printing at higher speeds and lower resolution for certain portions of the optical system (such as opto-mechanics and supporting structures) and with higher resolution for the portions of the system as required. Merging with electronic printers/magnetic material printers is another opportunity for hybrid modalities that could include a high level of photonic-electronic-actuation integration for all-in-one fabricated devices. An additional opportunity is in adding metrology feedback and incorporating tolerancing into the printing process. This could potentially bridge the gap between designer and fabrication facilities and enable complete prototyping in one setup. Toleranced 3D printing with metrology process control would provide easy-to-use optics or photonics fabrication tools to a broad research and development community and improve turn-around time during the prototyping and innovation process. An additional way in which 3D printing may impact optical design is by enabling the fabrication of non-symmetrical parts. Traditional methods of optical fabrication are limited to the fabrication of parts that are symmetrical in nature. Thus, 3D printing can enable the exploration of novel design concepts for optical elements. In any regard, additive manufacturing presents both opportunities and challenges but has a bright future ahead for inventors, researchers, and industry.