3D-printed immersion micro optics

Miniaturization can advance cutting edge optical applications in liquid environments^{1, 2} by reducing system size, reagent consumption and hence overall system cost^3-5. Available off-the-shelf optical components, however, such as macroscopic immersion objective lenses (Fig. 1a), are too expensive for single-use measurements and too large for non-destructive in situ medical diagnostics inside of small, confined organ cavities. 3D-printed micro optics using femtosecond 3D-printing⁶ have been emerging as a suitable alternative. They combine high optical performance with a compact footprint on the order of hundreds of microns^{7, 8}, which is ideal for fiber tip applications⁹.

Fig. 1  3D-printed immersion micro optics are highly sought after for in situ applications. a State-of-the-art macroscopic immersion objective lenses are about $ 100 \times $ larger than 3D-printed micro optics. The flexible 3D-printing approach enables swift design adaptation to match the RI of different immersion media. b Compact immersion optics enable applications in confined cavities, e.g. spectroscopy in small arteries. c They could also provide stable coupling between fibers and photonic integrated circuits by immersion into glue, active alignment, and subsequent fixing.

So far, 3D-printed micro optics remained mostly limited to gaseous or evacuated environments, or for solid immersion⁹. Many in situ applications, however, would benefit from a practical approach that enables 3D-printed micro optics that are viable for immersion into liquids.

In the biomedical context, successful lab-on-a-tip miniaturization requires immersion optics for illumination and to probe light that carries information about a liquid sample¹⁰. Raman-spectroscopy for instance excels in label-free identification of cells¹¹ for disease and tissue diagnostics¹² in vitro and in vivo^{12, 13}. Similarly, miniaturized fluorescence correlation spectroscopy may enable minimal invasive in situ analyses¹⁴ of dynamic biomolecular properties like diffusion rates, viscosity, binding constants or concentration changes¹⁵ inside of confined patient tissue and small blood vessels (Fig. 1b).

Photonic integrated circuits (PIC) and quantum dot applications encourage compact and stable coupling to optical fibers¹⁶. 3D-printed photonic wire bonds enable miniaturized and efficient chip-to-chip- and off-chip coupling for PICs^{17, 18}, but their in situ manufacturing requires extensive pre-alignment. Alternatively, lensed fibers, grating couplers¹⁹, tapered edge couplers²⁰, and 3D-printed micro optics^{21, 22} may provide compact coupling, but they are unstable due to free-space operation. More robust connections were pursued by 3D-printing of an alignment chuck directly onto a quantum dot to insert and glue an optical fiber^{23, 24}. Positioning such a bulky chuck in respect to the quantum emitter, however, was delicate and required extra markers. Immersion micro optics could simplify this process, as glue could be cured in the gaps between micro optical systems and photonic integrated circuits after coupling their modes (Fig. 1c).

An optical surface's ability to refract transmitted light depends on the refractive index (RI)-difference across the interface. Available photoresists for 3D-printing of micro optics ($ \text{n} \approx 1.5 $) are well suited for application in air ($ \text{n} \approx {1} $), but resemble the RI of liquid immersion media ($ \text{n} \geq 1.33 $) very closely. To achieve suitable focusing power, an immersed lens hence requires an extremely strong surface curvature, impractical for femtosecond 3D-printing. Diffractive optical elements (DOEs)^25-27 and metalenses^{28, 29} offer to circumvent this problem. Compared to the use in air, however, those devices require higher aspect ratios in immersion. This complicates manufacturing, exacerbates shadowing effects and reduces diffraction efficiency, especially when a high numerical aperture (NA) is desired³⁰.

Engineering a high-RI photoresist to increase the RI-contrast between the lens material and the immersion medium offers another approach towards high performance 3D-printed immersion micro optics. To estimate the required RI, a spherical surface with a constant curvature could be considered: For the same paraxial refractive power as in air ($ \text{n} \approx 1 $), immersion of the surface into water ($ \text{n} = 1.33 $) requires raising the lens material's RI from $ \text{n} = 1.5 $ to at least $ \text{n} = 2 $⁴². Commercial resists such as SU-8, IP-S, IP-Dip, and IP-n162, or widely used custom resists such as SZ2080$ ^{\text{TM}} $⁴³ have RIs well below this ideal threshold for immersion (Table 1). Doping or admixing nanoparticles can increase the resulting material's RI up to $ \text{n} \approx 1.9 $. Higher volume fractions, however, quickly cause fatal absorption and scattering artifacts. Thus, this approach's feasibility is limited to the engineering of photoresists with $ \text{n} < 2 $, which is too low even for cases with low-RI immersion media such as deionized water ($ \text{n} = {1.33} $).

Practical application for 3D-printed micro optics favors commercial photoresists with a good knowledge of critical material properties, including surface roughness⁴⁴, mechanical and physical behavior^45-47, and biocompatibility profiles⁴⁸. As a consequence, important routine practices like iterative surface shape correction to converge the fabricated lens’ surface to the designed shape^{49, 50} have been established for commercial photoresists. Experimental photoresists on the other hand typically lack comparable process knowledge and optimization.

Our 3D-printed immersion micro optical systems (3D-PRIMOPS) achieve high refractive power by maintaining an air-photoresist interface on internal optical surfaces via a microfluidic sealing technique. To the best of our knowledge, refractive 3D-printed liquid immersion optics either have been shielded with a protective sheath^{51, 52} or a bulky chuck^{23, 24} in a post fabrication assembly step. The required tolerances increase the systems’ mechanical footprint and make them susceptible to misalignment – limitations, that we avoid by combining monolithic manufacturing with a subsequent microfluidic self-sealing step to realize a protective encapsulation. In experiment, our manufactured 3D-PRIMOPS’ achieve diffraction limited performance with $ \text{NA} = 0.625 $, immersed in media ranging from water ($ \text{n} = {1.33} $) to oil ($ \text{n} = {1.52} $).

In a first step, ray optical simulations confirm that a refractive lens may achieve high refractive power and diffraction limited performance in air (Fig. 2a, case 1), but fails to focus incident rays upon immersion into water with $ \text{n} = 1.33 $ (case 2). Even if we bend the immersed lens far beyond the limit of measurement- and thus verifiable manufacturing capabilities (see Methods: Ray tracing and wave optical simulation), its refractive power remains insufficient to create a tight focus (case 3). We conclude that for powerful optical performance, we must avert immediate contact between lens and immersion medium by fabricating a cover structure for protective shielding (case 4) together with the lens, ideally as a single monolith.

Fig. 2  Our proposed air-tight encapsulation strategy provides micro optics for tight focusing in liquid immersion. a Ray optical simulation of light focusing by a 3D-printed lens. While capable of tight and diffraction limited focusing in air (case 1), immersion into water reduces the refractive power of the lens (case 2), preventing tight focusing despite re-optimization (case 3). Encapsulation prohibits intrusion of the immersion medium and preserves the optical system’s refractive power, enabling a lens with tight focusing capability (case 4). The white outlines indicate diffraction limited performance at the respective $ \text{NA} $ in wave optical simulations (Fast Polarized Wave Propagation Method, FPWPM)³¹. b Cylindric walls for encapsulation prohibit rinsing uncured photoresist from the capsule’s interior after the femtosecond 3D-printing process. c Microfluidic channels and vias facilitate creation of air-filled cavities. After loading by capillary action, the resin is cured by UV-light to permanently seal the vias.

While a cylindrical wall would geometrically describe a suitable 3D-encapsulation around the optical system (Fig. 2b), such an airtight structure would entrap uncured photoresist during manufacturing, failing to create an air-filled void. Therefore, we integrate microfluidic channels and vias into the encapsulating wall (Fig. 2c), akin to a process for creating opaque apertures⁵³. This facilitates solvation of entrapped resist during development to obtain an air-filled cavity. To prevent intrusion of immersion media through the same vias, we exploit the capillary effect to draw an UV-curable, viscous liquid into the microfluidic channels. Subsequent UV-curing seals all vias permanently.

Adopting this encapsulation strategy, we create a ray optical design of a 3D-PRIMOPS for point illumination, well suited for applications in spectroscopy or integrated fiber coupling. We aim for tight ($ \text{NA} = {0.625} $) and diffraction limited focusing into a silicone oil ($ \text{n} = {1.4} $) immersion medium (Fig. 3a). Behind the single mode fiber (SMF), a 650 μm long coreless fused silica fiber (CFSF), spliced at the facet of the SMF for mode expansion, and a subsequent 3D-printed doublet constitute the optical system.

Fig. 3  a Ray optical design of a point illumination 3D-printed immersion micro optical system (3D-PRIMOPS). Four surfaces, S1-S4, constitute the 3D-printed part. A vector wave optical simulation of the 3D-PRIMOPS predicts diffraction limited focusing, outlined by the white dashed line. CFSF: coreless fused silica fiber. b Light microscopy- and X-ray microtomography images of a manufactured, encapsulated 3D-PRIMOPS, clearly show the air-filled cavity and the seamless encapsulation.

The 3D-printed doublet comprises four interfaces (S1-S4, Fig. 3a). S1 and S2 build the 3D-PRIMOPS’ wall facing the CFSF, followed by an air-filled cavity, and behind that, S3 and S4 constitute the wall facing the immersion medium. All these surfaces exert both optical and mechanical functionality to achieve focusing and encapsulation in an assembly-free design. Because of the minute RI-contrast at both surfaces, S1 and S4 barely affect the optical performance. For robust adhesion to the optical fiber, we keep S1 flat, and to ensure robust wetting of S4 with the immersion media in this proof-of-concept work, we likewise keep S4 flat. The inner surfaces S2 and S3 exhibit high refractive power due to the large RI-contrast provided by the protective shielding. We therefore use these two surfaces to optimize a high-NA system with moderate angles of incidence on all surfaces, achieving a relaxed optical design. Especially noteworthy in our monolithic approach, S3 contributes to the optical performance positively, whereas in assembled approaches, this inner surface of the capsule usually causes considerable aberrations that require correction⁵². This avoids compromising tolerances and ensures high optical quality as well as robust manufacturing.

Since the small diameter ($ \approx 125\ {\text{μ}}{\rm{m}} $) of the optical system could cause disturbing diffraction effects, we perform a wave optical simulation for validation³¹ (Fig. 3a). The simulation confirms diffraction limited focusing for the ray tracing derived 3D-PRIMOPS.

Likewise, we optimize two more 3D-PRIMOPS' adjusted for water ($ \text{n} = {1.33} $) and oil ($ \text{n} = {1.52} $) immersion, to account for a broader range of immersion media. A separate 3D-PRIMOPS has to be manufactured for each distinct immersion medium. Since femtosecond 3D-printing allows for rapid manufacturing of individual surface shapes, the additional effort to adjust the design to different immersion media is minute.

For experimental proof, we manufacture the three 3D-PRIMOPS’ designed for water-, silicone-, and oil immersion with a femtosecond 3D-printer and carry out the microfluidic sealing process as outlined above (Fig. 2c). The total probe diameter at the fiber tip is as low as $ 205\ {\text{μ}}{\rm{m}} $. Light microscopy- and X-ray microtomography⁵⁴ images show the tightly sealed, air-filled cavity of the 3D-PRIMOPS (Fig. 3b). In experiment, we quantify the point spread function (PSF) of our manufactured 3D-PRIMOPS’ (Fig. 4a) and observe diffraction limited focusing for all three 3D-PRIMOPS’ (Fig. 4b).

Fig. 4  a Setup for PSF measurement of 3D-PRIMOPS’ in immersion. Light is coupled into the SMF with the 3D-PRIMOPS, which is immersed directly. A microscope setup images the 3D-PRIMOPS’ focal area onto a CMOS sensor. Volume stacks of the intensity in the 3D-PRIMOPS’ focal area are acquired using a piezo stepper for z-scanning. b Measured PSFs for water-, silicone- and oil immersion. The white outlines indicate diffraction limited performance at $ \text{NA} = {0.625} $.

Our results pave the way for manufacturing of complex refractive 3D-printed immersion micro optics with high optical performance and minimal mechanical footprint. The current optical design requires an array of 3D-printed lenses to mimic an immersion correction collar’s functionality (Fig. 1a); future optical designs, that are invariant to RI-changes of the immersion medium⁵⁵, could provide similar functionality within one single optical device to further minimize the fabrication time and the mechanical footprint. Moreover, the presented microfluidic shielding may promote various complex immersion systems for illumination, imaging, and light collection, such as stacked 3D-printed dielectric metasurfaces²⁸ or high-NA hybrid microlenses for holographic optical trapping⁵⁶. We anticipate broad applicability in miniaturized analytical devices for biomedical research, keyhole access medical endoscopy, and photonic integrated circuit coupling.