120 tiny beams leave a fiber. How do you bring more of their light into one spot? This paper adds a miniature, 3D printed optical system to the tip.

120 tiny beams leave a fiber. How do you bring more of their light into one spot? This paper adds a miniature, 3D printed optical system to the tip.[1]

A 3-minute exploration of Sivankutty, Bertoncini & colleagues (2022)

Original schematic · not to scale · based on Fig. 1

Mind the gaps.

Start with perfectly aligned phases. Then widen the beams. Watch the light outside the central peak: phase alignment alone cannot erase a sparse pupil.[2]

Build a better pupil

The pupil is the light pattern just before focusing. Try these three starting points.

0.10 = narrow beams. 1.00 = neighboring beams touch. Total model power stays fixed.

RMS phase error in radians. Zero means all beams are in step. Arrow directions show each beam’s phase.

Horizontal: dimensionless spatial frequency q. Shading marks |q| ≤ 1/9, the aligned array’s central lobe.

The phases agree, but the pupil has gaps. Light still reaches satellite peaks. Now choose “Wider & aligned” to concentrate the diffraction pattern.

This is a scalar, one-dimensional Fraunhofer demonstration with nine equally spaced rectangular beams of equal power. The paper uses 120 cores in an aperiodic Fermat spiral, Gaussian beams, real micro-optics and a non-paraxial wave-optics model. These sliders are conceptual controls; they do not specify lens dimensions or predict device performance. The regular layout here makes satellite peaks easy to see. In the paper’s aperiodic layout, energy outside the focus is redistributed rather than forming the same regular peaks (p. 3).

For beam width / spacing f, spatial frequency q and beam phase φₙ, the plotted intensity is normalized to the on-axis peak of a fully filled, aligned pupil at the same total power:

I(q) / I₍ideal₎(0) = (f / N²) sinc²(fq) |Σₙ exp[iφₙ − i2πqn]|² N = 9; n = −4,…,4; sinc(x) = sin(πx)/(πx).

The phase slider scales a fixed, zero-mean set of phases to the specified RMS error. “Phase coherence” is |Σ exp(iφₙ)|² / N². On-axis intensity is f times that value. Neither metric is the paper’s power efficiency η, which integrates power over a focal spot. The plot displays only −3 ≤ q ≤ 3; light outside this window is not shown. The pupil drawing shows width and phase, not the amplitude increase needed to keep narrow beams at fixed power.

Assumptions: monochromatic scalar light; ideal lenses; no polarization, clipping losses, aberrations, tissue, pulse dynamics or nonlinear fluorescence. This demonstration illustrates the Fourier principle in Eqs. (1)–(2), p. 3. It does not reproduce the paper.

Expand. Align. Focus.

Keep cores separated

Widely spaced fiber cores help limit cross-talk, but leave gaps in the emerging light. Even phase-aligned beams send energy outside the central focus.[2]

Fill the optical pupil

A lenslet over each core expands and collimates its beam. A shared lens brings the expanded beams together. The lenslets change the light pattern without moving the cores.[1]

Bring phases into step

A spatial light modulator compensates the relative core phases. Constructive interference then forms the focus; the printed optics improve how much power reaches it.[3]

More of the light. In the right place.

Combining efficiency η is the power in the central focal spot divided by the total power. These are reported values, separate from the workbench above.[3]

The reported ~35-fold improvement compares with earlier unstructured fibers; it is not a new paired control experiment shown here. The measured 35% is 70% of the modeled 50%.[3]

25 mW at the tip. About 9 mW in focus.

The authors imaged 2 µm fluorescent beads with 150 fs pulses at 920 nm. This demonstrates compatibility with nonlinear imaging, where a concentrated spatiotemporal focus matters. It is a bead experiment, not a tissue or clinical demonstration.[4]

A promising device. A bounded result.

The paper demonstrates a miniature beam combiner and bead imaging. It does not establish performance in living tissue.

Efficiency has a measurement boundary. The 35% characterization uses a 1030 nm continuous-wave laser and selects s-polarized light. The fiber does not preserve polarization; 35% is not established for all output polarizations.[3]

Real optics fall short of the ideal. The authors primarily attribute the 35% versus 50% gap to top-lens field curvature and reduced beam overlap. The measured spot is about 8% wider than their ideal design.[3]

Imaging scope is limited. The study uses a 200 mm fiber and fluorescent beads. Figure 4 reports 100 ms per pixel, limited by the SLM update rate. This experiment does not demonstrate video-rate tissue imaging.[4]

The source paper

Full eight-page preprint read. arXiv:2203.01636v1 · 3 March 2022 Scientific review required.

Miniature 120-beam coherent combiner with 3D printed optics for multicore fiber based endoscopy

Siddharth Sivankutty, Andrea Bertoncini, Victor Tsvirkun, Naveen Gajendra Kumar, Gaelle Brévalle, Géraud Bouwmans, Esben Ravn Andresen, Carlo Liberale & Hervé Rigneault.

All diagrams here are original. No source figures are reproduced. Page references use the eight-page arXiv PDF. Text extraction preserves all pages, but some PDF mathematical glyphs may not render faithfully; consult the PDF for notation.

Device and optics: pp. 2–4, Fig. 1 and design/fabrication discussion; pp. 6–7, conclusion. 120 beams; 180 µm diameter, 200 µm height; microlenses expand the beamlets and a shared lens focuses them. Design: 39 µm lenslet focal length, 800 µm top-lens focal length, 100 µm lens separation. PDF p. 4 ↗

Why fill factor matters: pp. 1–3, especially p. 3, Eqs. (1)–(2) and the following Fourier argument. Sparse near-field amplitude sends energy into nonzero spatial frequencies even after co-phasing. Aperiodicity redistributes this energy; it does not fix focusing efficiency. PDF p. 3 ↗

Modeled and measured performance: p. 4, ideal design η = 0.5, spot 4.8 µm, FOV about 60 µm; pp. 5–6, characterization and Fig. 3, 1030 nm CW source, s-polarization selection, η = 0.35, quantitative spot size 5.2 µm FWHM, earlier unstructured efficiency about 0.01 or lower, and discussion of field curvature. The text also rounds the spot to 5 µm in its Fig. 3 discussion. PDF p. 6 ↗

Two-photon imaging and scanning: p. 6, imaging paragraph; p. 7, Fig. 4 and caption. 150 fs / 920 nm excitation, 200 mm fiber, 56 µm FOV at half-max efficiency, 2 µm beads, 25 mW total and about 9 mW focused, 5 µm lateral / 60 µm axial resolution. Figure 4: 48 × 48 µm image field, 20 µm plane spacing, 100 ms/pixel. PDF p. 7 ↗