3D-Structured Polyethylene Windows for Low-Loss Transmission in Wideband Cryogenic Terahertz Systems

TL;DR

The paper tackles the challenge of designing broadband, low-loss cryogenic vacuum windows and infrared filters for wideband THz receivers by engineering pyramidal AR textures directly into UHMWPE. It combines full-wave FEM modeling and EMT for rapid prototyping with CNC fabrication to realize a robust 3D-structured surface, and validates performance with THz-TDS and heterodyne measurements. Results show transmission of $97-99%$ across $211-373 GHz$ (extending to 500 GHz) and minimal added noise ($2-12 K$) in practical receiver configurations, with polarization preserved. The study demonstrates that structured UHMWPE is a practical, scalable platform for wideband, cryogenic optics in high-sensitivity THz instrumentation, and it highlights potential refinements to further enhance throughput and thermal performance.

Abstract

We present the design, fabrication, and characterisation of a broadband vacuum window and infrared filter based on ultra-high molecular weight polyethylene (UHMWPE) for millimeter wave receivers operating across ALMA Band 6 and 7 (211-373 GHz). The window incorporates pyramidal anti-reflection (AR) structures, machined directly into the polyethylene using CNC machining, which provide impedance matching over a broad frequency range. The structured UHMWPE method was implemented in two distinct components: a vacuum window and a cryogenic infrared filter. This surface structuring approach provides mechanical robustness, cryogenic compatibility, and low insertion loss. We characterize the transmission properties using Terahertz Time-Domain Spectroscopy (THz-TDS), which demonstrates reflection below $5\%$ across the full band. Complementary heterodyne measurements confirm improved receiver noise performance. These results establish 3D structured UHMWPE as a promising platform for broadband cryogenic optics in high-sensitivity THz instrumentation.

Figures (10)

• Figure 1: Illustration of the UHMWPE vacuum window prototype.
• Figure 2: Co‑polar performance of two broadband AR unit cells at normal incidence. (a) Transmission Tx and (b) reflection Rx for a pyramidal profile and a three‑step groove profile. Curves for both incident linear polarizations (TE: E$\parallel$grooves,TM: E$\perp$grooves) are overlaid; co‑polar transmission/reflection are indistinguishable at normal incidence. Both unit cells share the same pitch (p) and depth (h); the three‑step layer heights/widths were chosen to approximate the triangular effective‑index gradient.
• Figure 3: Comparison of full‑wave (FEM) and effective‑medium (EMT) models for the UHMWPE pyramidal AR window and parametric sensitivity of the mean transmission. TE and TM polarisation are perfectly overlaid. (a) Normal‑incidence co‑polar transmission T(f) from 211–373 GHz for a single unit cell solved with FEM (periodic boundaries, Floquet ports) and with EMT (staircase discretization of the graded profile and transfer‑matrix propagation). (b) Corresponding co‑polar reflection R(f) over the same band. (c) Band‑averaged transmission $\hbox{$\m@th T$}{ \hbox{$\m@th\overline{\hbox{$\m@th\overline{\hbox{$\m@th\overline{}$}}$}}$}} _{211-373 GHz}$ versus apex angle $\alpha$ at fixed pitch p=0.5mm. (d) Band‑averaged transmission versus pitch p at fixed $\alpha = 22^{\circ}$. The dashed lines indicate the nominal design values used for fabrication of the AR structure.
• Figure 4: Simulated transmission and reflection coefficients of the structured UHMWPE vacuum window at normal incidence, computed via FEM. Two geometries are compared: an idealized profile with sharp pyramidal tips and no flat regions, and the fabricated structure based on experimentally measured groove dimensions, which includes flat-tipped pyramids and inter-groove spacing. (a) Co-polarized and cross-polarized transmission for incident electric fields aligned with the groove direction on the front surface. Co-polar transmission corresponds to detection with an output port aligned with the incident polarization; cross-polar transmission is measured using a waveguide port rotated by 90°, measuring polarization rotation induced by the structure. (b) Co-polarized reflection. (c) Cross-polarized reflection. For all the plots, TE and TM polarisation are perfectly overlaid.
• Figure 5: a) Top-view optical image of the machined pyramidal corrugation on the UHMWPE window, showing the uniform two-dimensional array of grooves. (b)Optical-profilometer cross-section cut through adjacent grooves, showing a valley flat width of approximately 50 $\mu m$, a truncated pyramid apex flat of 150 $\mu m$, and an apex angle of 15$^\circ$. (c) Three-dimensional optical-profilometer rendering of the corrugation, showing the uniform facet slopes.