Vertical drawing system sculpts microfibers for tailored supercontinuum light

Optical micro/nanofibers owe their most useful properties to geometry. When a silica fiber is drawn down to a diameter comparable to the wavelength of the light traveling through it, a large fraction of the optical field leaks into an evanescent tail outside the glass, the field becomes tightly confined, and both dispersion and the nonlinear coefficient become strongly dependent on diameter. That diameter dependence is an opportunity: shape the fiber’s profile along its length and you shape the nonlinear physics inside it. A team reporting in Frontiers of Optoelectronics has now built a drawing system that attacks the problem from an unexpected direction, literally, by pulling fibers vertically rather than horizontally, and has used it to fabricate a diameter-gradient microfiber with four cascaded structures over roughly 120 millimeters and a minimum diameter of about 1 micrometer.

The motivation for the vertical arrangement comes from a familiar laboratory nuisance. In conventional horizontal tapering systems, a flame, an electric heater, or a laser softens the fiber while simultaneously heating the surrounding air, and hot air rises. For fibers with simple geometries, the structural strength of the glass resists this upward airflow, which is why every previously reported fabrication approach used horizontal processing. But when a fiber must pass through multiple sequential drawing steps to build a complex structure, the airflow can induce asymmetrical bending that becomes permanently frozen into the fiber during the final annealing stage, since the deformation occurs while the glass is hot and soft. The researchers illustrated this with structural and stress simulations in ANSYS Workbench, using the measured temperature distribution of their heating zone. A conventional fiber bent symmetrically upward under the airflow, while a simulated intermediate fiber with two cascaded structures, whose lateral geometry is asymmetric, experienced a more complex stress distribution and greater deformation, including horizontal asymmetry on top of the upward bending.

The new system counters the airflow with two strategies. First, the home-made heating element, built from ceramics and silicon nitride heating elements bonded with high-temperature structural adhesive, is enclosed in a ceramic casing that insulates the heating zone and reduces air convection. Second, and more fundamentally, the drawing configuration is vertical, so the direction of thermal convection is parallel to the fiber rather than perpendicular to it. The heater forms a cavity of approximately 4 by 4 by 40 millimeters through which the fiber passes, with temperature measured in real time by a type-B thermocouple and controlled by a temperature controller. The uniform temperature in the heating zone was 1157 degrees Celsius, with lateral temperature variation of less than 0.1 percent within 4 millimeters. The fiber is clamped at both ends and pulled by two high-precision translation stages driven by a motion control card and a servo, following preset drawing programs that support constant hot-zone, linear hot-zone variation, and scanning simulated constant hot-zone methods. The heaters last roughly 1000 hours under standard operating conditions and cost below 21 US dollars each, keeping the whole apparatus a table-top instrument rather than a fiber-drawing tower.

Industrial fiber drawing has long proceeded vertically, in the molten core approach that produces uniform cross-sections over kilometer-scale lengths, but the towers required for that process stand tens of meters high and are costly to build and operate. The new system borrows the vertical principle while shrinking it to laboratory scale. To demonstrate what it can do, the team drew a standard 125-micrometer fiber first into a section approximately 45 millimeters long and 4.5 micrometers in diameter using the linear hot-zone variation method, then used the constant hot-zone method in four further steps to draw four tapered regions, each 8 millimeters long and 1.3 micrometers in diameter. The finished fiber was characterized by stitching together 850 microscope images taken with a 20-times objective lens, yielding a maximum diameter of 4.5 micrometers and a minimum of 1.3 micrometers, closely matching the designed specifications.

With the fabrication tool in hand, the researchers turned to the design question: what fiber geometry produces the flattest, widest supercontinuum? Supercontinuum generation arises from the combined action of several nonlinear effects, including soliton fission, Raman self-frequency shift, and Cherenkov radiation from dispersive waves, so optimizing the structure by hand would require a large number of attempts. Instead, the team solved the generalized nonlinear Schrödinger equation numerically, using a fourth-order Runge–Kutta method built on the split-step Fourier technique with an adaptive step-size strategy to reduce computation time, and defined a flatness index based on the standard deviation of spectral intensity relative to its average. They then applied a genetic algorithm to reverse-design the fiber structure for a target spectrum under defined pumping conditions.

The simulations mapped how each parameter shapes the output. Decreasing the taper slope broadened the simulated spectrum from 604.44 nanometers to 826.86 nanometers at the 35-decibel level, but with trade-offs in both directions: a slope that is too large shifts the second zero-dispersion wavelength too rapidly along the fiber, preventing energy transfer beyond it, while a slope that is too shallow extends the taper length, reduces peak power, and makes fabrication harder. Increasing the number of cascaded structures from zero to five expanded the simulated bandwidth from 151.28 nanometers to around 612.70 nanometers, with more prominent higher-order soliton fission in the four- and five-segment configurations. Raising the pulse width from 0.08 to 0.15 picoseconds broadened the spectrum from 624.10 to 723.12 nanometers, and increasing peak power from 800 to 2000 watts expanded it from 276.43 to 671.72 nanometers, because higher power shortens the nonlinear length and raises the soliton order. The genetic algorithm ultimately returned an initial waist diameter of 4.5 micrometers with a length of 42 millimeters, a secondary taper waist diameter of 1.32 micrometers with a length of 8 millimeters, and four cascades.

Before the optical tests, the team verified the drawn fibers’ dispersion using white light interferometry, a setup in which light from a broadband amplified spontaneous emission source is split by a 3-decibel coupler into two paths, one containing an ordinary single-mode fiber of known dispersion and the other the microfiber under test. By adjusting the single-mode fiber length to optimize interference fringe contrast and extracting phase information from the fringes, they reconstructed the group velocity dispersion of fibers ranging from 1.2 to 2 micrometers in diameter, with experimental and simulated results deviating within acceptable limits. This step matters because supercontinuum generation and dispersion are closely linked, and the measurements allowed the simulations to be fitted against the actual fibers rather than idealized models.

The optical experiments used a femtosecond erbium-fiber laser producing pulses of approximately 120 femtoseconds at a central wavelength of 1550 nanometers, coupled into the microfiber through an achromatic collimator after power adjustment with a neutral density filter, with the output spectrum recorded on an optical spectrum analyzer. Coupling 1050-watt peak-power pulses of 0.12-picosecond width into the optimized cascaded fiber produced a supercontinuum spanning 1463 to 1741 nanometers at the minus-10-decibel level, an output bandwidth of 277.85 nanometers with a flatness of approximately 0.4874, and an efficiency of 264.62 nanometers per kilowatt of peak pump power. The comparison experiments bore out the simulation conclusions. As the number of cascades increased from one to four, the measured bandwidths at the 10-decibel level were 62.70, 201.25, 173.47, and 277.85 nanometers. A fiber with a larger taper slope, drawn with a laser heating system that provides a shorter heating zone, managed only 215.52 nanometers of bandwidth with a flatness of 0.5835, confirming that the long, gently sloped taper promotes better phase matching of dispersive waves and four-wave mixing. And at 850 watts the optimized structure delivered 231.14 nanometers with a flatness of 0.5823, so the higher pump power improved both width and flatness simultaneously.

By the efficiency metric the authors defined, bandwidth in nanometers divided by peak pump power in kilowatts, their cascaded structure achieved the highest supercontinuum generation efficiency among the previously reported fiber structures they tabulated. The cascaded approach also avoids the fusion splicing that traditional cascade fabrication requires, which typically leaves longer tail fibers and imposes restrictions on fiber structure; the multi-step tapering demonstrated here is more compact and freer in geometry. The authors note that the method is not limited to the 1550-nanometer band and could be applied at other wavelengths such as 850 and 1030 nanometers. As with any laboratory demonstration, the results rest on a single optimized fiber design under specific pump conditions, and the reported efficiency comparison depends on the structures included in the authors’ table. Still, the underlying message is straightforward: controlling the direction of hot air, and the geometry of the glass, turns a precision fabrication problem into a design space for nonlinear optics, and a table-top instrument may now be enough to explore it.

Subject of Research: A vertical micro-nanofiber drawing system for fabricating cascaded diameter-gradient fibers that generate supercontinuum spectra

Article Title: Precision vertical drawing of diameter-gradient microfibers: cascaded geometries for tailored nonlinearity

Article References: Chi, H., He, X., Lu, D., Wang, S., Wu, J., Jin, M., Li, X., Wang, Z., & Ma, Y. (2025). Precision vertical drawing of diameter-gradient microfibers: cascaded geometries for tailored nonlinearity. Frontiers of Optoelectronics, 18(1), Article 16. https://doi.org/10.1007/s12200-025-00160-8

Image Credits: AI Generated

DOI: 10.1007/s12200-025-00160-8

Keywords: micro-nanofibers, fiber drawing, vertical drawing system, supercontinuum generation, nonlinear optics, thermal airflow, taper slope, genetic algorithm optimization, dispersion measurement, silica fibers, photonics, spectrum broadening