Ultra-broadband wireless rectification and frequency mixing via the nonlinear Hall effect in TaIrTe4 Authors: Fanrui Hu, Jiayu Lei, Shishun Zhao, Qu Yang, Nils Dessmann, Thom Janssen, Chenhui Zhang, Hao Yu, Peng Yu, Andrei Kirilyuk, Hyunsoo Yang Published: 2026 ABSTRACT Driven by the advancements in the Internet of Things and emerging 6 G technologies, the demand for broadband energy harvesting and frequency conversion is surging. Traditional semiconductor devices face fundamental challenges at high frequencies, particularly in the terahertz range. Here, we demonstrate an ultra-broadband rectifier and a passive mixer based on the nonlinear Hall effect (NLHE) in a Weyl semimetal TaIrTe4. The device achieves rectification across a broad range, from 19 MHz to 2.88 THz, at room temperature, operating without electrical bias or an external magnetic field. By employing series connections, we enhance the rectified voltage and accomplish polarization-insensitive operation. Furthermore, we realize a broadband (0.1–40 GHz) passive mixer with a conversion gain of − 20 dB and high linearity (IP1dB = 4 dBm) based on NLHE in TaIrTe4. These results establish NLHE in quantum materials as a powerful platform for next-generation high-frequency electronics. Introduction The rapid expansion of the Internet of Things (IoT), wearables, and portable electronics has created a pressing need for sustainable power solutions, especially in environments where frequent battery replacement or recharging is impractical1–3. With billions of smart devices deployed globally, developing self-powered energy sources is critical for their continuous operation4. 5 G and emerging 6 G technologies generate abundant ambient electromagnetic (EM) waves across a wide spectrum, from lower GHz to millimeter-wave (mmWave) and terahertz (THz) bands5–7. Without effective energy conversion mechanisms, this EM energy remains largely untapped. Broadband energy harvesters can convert EM waves across diverse frequency bands into usable electrical power, offering a promising solution for sustainable and portable power sources while optimizing the use of ambient EM energy. The core of an energy harvester is a rectenna, which combines an antenna to capture EM energy and a rectifier to convert the EM wave into direct current (DC). Broadband rectifiers have been developed using nonlinear components, such as Schottky diodes8,9, field-effect transistors (FETs)10–13, and ballistic rectifiers14. Advanced broadband rectifiers based on Schottky junctions, including those utilizing two-dimensional MoS2, have demonstrated continuous operation from 0.1 GHz to 10 GHz8. However, extending their efficient operation into the THz range, which is vital for future communication bands, remains challenging due to the parasitic capacitance at higher frequencies15. Early CMOS FET rectifiers exhibited broadband responsivity from 0.27 to 1.05 THz10, while more recent FET devices have demonstrated rectification at approximately 3 THz13. Similarly, graphene ballistic rectifiers have also achieved THz broadband rectification from 0.07 to 0.69 THz14. Despite the performance in the THz range, these rectifiers are limited in harvesting energy from commonly used lower frequencies, such as 2.4 GHz for Wi-Fi. Therefore, developing a rectifier with energy-harvesting capability that covers MHz to THz frequencies is valuable for advancing self-powered IoT devices, 6 G networks, and other emerging technologies. Frequency mixers are another critical high-frequency component in modern communication systems, utilizing nonlinear components to generate sum and difference frequencies for signal modulation and demodulation16. Mixer designs primarily rely on CMOS and III-V semiconductor technologies16,17. However, achieving high linearity and broadband operation remains challenging16,18. In addition, the fabrication complexity of CMOS mixers increases at higher frequencies, as deep-submicron scaling introduces difficulties such as increased flicker noise, reduced voltage headroom, and weaker port-to-port isolation16,18. The nonlinear Hall effect (NLHE) has recently emerged as a promising mechanism for developing rectifiers and mixers, offering a simple, junction-free design for high-frequency applications19–22. In NLHE, an alternating EM field generates a transverse Hall voltage without requiring an external magnetic field. This Hall voltage contains both a DC component for rectification and higher-order harmonics for frequency mixing. Compared with conventional junction-based technologies, NLHE primarily arises from intrinsic properties of quantum materials, such as the Berry curvature dipole (BCD)20,23 or quantum metric dipole24,25, which enable signal modulation without junctions at high-frequency19,26,27. Extrinsic mechanisms may also contribute to NLHE28–30. Recent studies have shown NLHE-induced rectification in specific frequency regimes, including the GHz21,24,31–34 and the THz frequencies35,36, each within a restricted spectral window. Mixers driven by NLHE have so far lacked systematic evaluations of key performance metrics such as operational bandwidth and linearity22,36. Here, we show an ultrabroadband nonlinear Hall rectifier operating from tens of MHz to several THz, enabling a single-device platform for broadband wireless signal conversion. The scalability of the rectifier through array integration highlights its potential for wireless energy harvesting capability. Beyond rectification, we quantitatively evaluate key parameters for frequency mixing, underscoring the potential of NLHE for broadband high-frequency electronics. Results NLHE in TaIrTe4 Our experiment is conducted in Td-phase TaIrTe4, a type-II Weyl semimetal. The crystal structure of TaIrTe4, depicted in Fig. 1a, belongs to the noncentrosymmetric orthorhombic space group (Pmn21)37,38. The elemental components of the material are identified by energy dispersive X-ray spectroscopy (Supplementary Note 1). The broken inversion symmetry in this material separates the positive and negative Berry curvatures in different momentum regions, generating a significant BCD20,21,39, which drives the intrinsic NLHE. To confirm the absence of inversion symmetry, we perform angular-resolved Raman spectroscopy and resistance measurements on TaIrTe4 flakes (Supplementary Note 2). The Raman characterization results in Fig. 1b reveal the asymmetric inversion and confirm the crystallographic alignment40,41.Fig. 1Crystal characterization and NLHE in TaIrTe4.a Crystal structure of TaIrTe4. The mirror symmetry is preserved in the b–c plane but absent in the a–c plane. The dashed line represents the mirror plane. b Angular-resolved Raman spectra of TaIrTe4 by rotating the sample with respect to the laser polarization direction. c Second-harmonic transverse (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{2\omega }$$\end{document}Vb−aa2ω) and longitudinal (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{aa}}^{2\omega }$$\end{document}Va−aa2ω) voltages as a function of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω in a 16 nm TaIrTe4 sample. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω is an AC with ω = 13.7 Hz, which is applied along the a axis. d Second-harmonic transverse \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{bb}}^{2\omega }$$\end{document}Va−bb2ω and longitudinal \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{bb}}^{2\omega }$$\end{document}Vb−bb2ω voltages as a function of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω in the same 16 nm TaIrTe4. The current is applied along the b-axis. The error bars in c and d reflect the standard deviation of five measurements. Crystal characterization and NLHE in TaIrTe4. a Crystal structure of TaIrTe4. The mirror symmetry is preserved in the b–c plane but absent in the a–c plane. The dashed line represents the mirror plane. b Angular-resolved Raman spectra of TaIrTe4 by rotating the sample with respect to the laser polarization direction. c Second-harmonic transverse (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{2\omega }$$\end{document}Vb−aa2ω) and longitudinal (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{aa}}^{2\omega }$$\end{document}Va−aa2ω) voltages as a function of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω in a 16 nm TaIrTe4 sample. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω is an AC with ω = 13.7 Hz, which is applied along the a axis. d Second-harmonic transverse \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{bb}}^{2\omega }$$\end{document}Va−bb2ω and longitudinal \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{bb}}^{2\omega }$$\end{document}Vb−bb2ω voltages as a function of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω in the same 16 nm TaIrTe4. The current is applied along the b-axis. The error bars in c and d reflect the standard deviation of five measurements. We fabricate the device by exfoliating TaIrTe4 flakes onto Si/SiO2 substrates (Methods). The electrodes are patterned along the crystallographic a and b directions of TaIrTe4 for transport measurements. The intrinsic NLHE in TaIrTe4 generates a second-order Hall voltage along the b axis when an alternating current (AC) is applied along the a axis, parallel to the BCD in TaIrTe4. To probe the NLHE, we apply an AC (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega },$$\end{document}Iω,ω = 13.7 Hz) to a 16 nm TaIrTe4 device and measure both the first (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{\omega }$$\end{document}Vω) and the second (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{2\omega }$$\end{document}V2ω) harmonic voltage by lock-in amplifiers at room temperature (Methods). When the AC is injected along the a axis of TaIrTe4, we observe a prominent second harmonic transverse voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{2\omega }$$\end{document}Vb−aa2ω that scales quadratically with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I}^{\omega }$$\end{document}Iω, while the corresponding longitudinal signal (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{aa}}^{2\omega }$$\end{document}Va−aa2ω) remains negligible (Fig. 1c). The negligible first harmonic transverse voltage confirms fine Hall bar alignment (Supplementary Note 3). To further validate the NLHE signal, we apply the AC along the b axis and measure the second-harmonic voltage along both a (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{bb}}^{2\omega }$$\end{document}Va−bb2ω) and b (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{bb}}^{2\omega }$$\end{document}Vb−bb2ω) axes. As expected, no discernible second harmonic signal is detected in either direction (Fig. 1d). These observations confirm the symmetry and directional dependence of the BCD in TaIrTe4, and establish a robust basis for exploring its rectification and mixing behavior. Wireless rectification in a TaIrTe4 rectifier The nonlinearity in NLHE provides a foundation for rectification and frequency mixing (Supplementary Note 4). After validating NLHE in TaIrTe4, we investigate the rectification of the device using a wireless input (Supplementary Note 5). Figure 2a shows the experimental setup, where a directional EM wave E(ω) is irradiated on the TaIrTe4 rectifier wirelessly through a linearly polarized antenna. The frequency 5.9 GHz is selected because it is a typical band of the latest Wi-Fi and vehicular communication42. The input power Pin is defined as the power directly emitted from the signal source, without correcting for power received by the device to estimate the actual front-end performance (Methods). Electrodes are aligned along the a and b axes to measure the rectified voltage, while the angle θ is the polarization direction of the EM wave relative to the a axis of TaIrTe4. Notably, we observe a polarization dependent rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b}^{{DC}}$$\end{document}VbDC with an input power Pin of 5 mW along the b axis of TaIrTe4, with a maximum response of 0.12 mV (Fig. 2b). The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b}^{{DC}}$$\end{document}VbDC is maximized when the EM wave aligns with the a axis (θ = 0° or 180°) and minimized when the EM wave is along the b axis (θ = 90° or 270°), consistent with the symmetry constraints of NLHE in TaIrTe4. In contrast, no rectified signal is detected along the a axis (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a}^{{DC}}$$\end{document}VaDC) at any polarization angle, as the mirror symmetry forbids NLHE in this direction. As control experiments, similar polarisation-dependent measurements are conducted on sputtered Pt and Cu devices, which lack broken inversion symmetry and thus cannot exhibit NLHE. There is no observable signal in Pt and Cu devices (Supplementary Note 6).Fig. 2Wireless rectification based on NLHE in TaIrTe4.a Schematic illustration of the wireless rectification experiment. The alternating EM wave (EAC) is emitted to the TaIrTe4 rectifier through a linearly polarized broadband antenna. b The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{{DC}}$$\end{document}VDC as a function of θ in the TaIrTe4 sample. The angle θ is defined as the polarization direction of the EM wave with respect to the a axis of TaIrTe4. c \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{{DC}}$$\end{document}VDC as a function of input power Pin at 5.9 GHz. Rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC is measured along the b axis when the EM wave is aligned along the a axis. The control experiment \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{aa}}^{{DC}}$$\end{document}Va−aaDC is measured along the a axis when the EM wave is aligned along the a axis. The optical image denotes the TaIrTe4 device, with a scale bar of 4 μm. The TaIrTe4 flake has an area of approximately 77 μm2, with a channel length (distance between electrodes) of 3.6 μm. d, e The polarization dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC across 110–170 GHz (d) and 170–260 GHz (e). f The rectified power \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{DC}}$$\end{document}Pb−aaDC as a function of load resistance RL measured at 158 GHz with the input power of 5 mW. The inset presents a schematic diagram illustrating the integration of the load into the circuit. The optical image in the top-right corner shows the TaIrTe4 device under test, with a scale bar of 4 μm. The data in (b–e) are collected from a 16 nm thick TaIrTe4 device. The data in (f) are measured from a 22 nm thick TaIrTe4 device. The error bars in b, c, and f reflect the standard deviation of five measurements. Wireless rectification based on NLHE in TaIrTe4. a Schematic illustration of the wireless rectification experiment. The alternating EM wave (EAC) is emitted to the TaIrTe4 rectifier through a linearly polarized broadband antenna. b The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{{DC}}$$\end{document}VDC as a function of θ in the TaIrTe4 sample. The angle θ is defined as the polarization direction of the EM wave with respect to the a axis of TaIrTe4. c \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{{DC}}$$\end{document}VDC as a function of input power Pin at 5.9 GHz. Rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC is measured along the b axis when the EM wave is aligned along the a axis. The control experiment \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{a-{aa}}^{{DC}}$$\end{document}Va−aaDC is measured along the a axis when the EM wave is aligned along the a axis. The optical image denotes the TaIrTe4 device, with a scale bar of 4 μm. The TaIrTe4 flake has an area of approximately 77 μm2, with a channel length (distance between electrodes) of 3.6 μm. d, e The polarization dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC across 110–170 GHz (d) and 170–260 GHz (e). f The rectified power \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{DC}}$$\end{document}Pb−aaDC as a function of load resistance RL measured at 158 GHz with the input power of 5 mW. The inset presents a schematic diagram illustrating the integration of the load into the circuit. The optical image in the top-right corner shows the TaIrTe4 device under test, with a scale bar of 4 μm. The data in (b–e) are collected from a 16 nm thick TaIrTe4 device. The data in (f) are measured from a 22 nm thick TaIrTe4 device. The error bars in b, c, and f reflect the standard deviation of five measurements. We further examine the input power dependence of the rectification signal using an EM wave at 5.9 GHz. The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC increases linearly with the input power Pin, following the relationship \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC ∝ \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{{{\rm{in}}}}^{\beta }$$\end{document}Pinβ, with β = 0.988, as shown in Fig. 2c. The slight deviation of β from unity can be attributed to experimental setup, such as reflection and free-space propagation losses between the antenna and the device. The linear behavior aligns with theoretical expectations, as the rectified voltage generated by NLHE is proportional to the square of the incident EM field, VDC ∝ E(ω)2 (Supplementary Note 7), and the input power scales with the square of the EM field (Pin ∝ E(ω)2). This agreement confirms that the rectification signal originates from the NLHE in TaIrTe4. Linear power dependence is consistently observed at 345 MHz, 765 MHz, 18 GHz, and 31 GHz (Supplementary Note 8), further validating the power-dependent relationship. Building on the rectification observed at lower GHz bands, we further investigate the broadband response at sub-THz frequencies (over 100 GHz). This frequency regime is considered pivotal for the next decade due to its wide applications in 6 G communication, advanced sensing and security technologies43–45. During the experiment, we irradiate the rectifier with EM waves in the waveguide D band (110–170 GHz) and WR4 band (170–260 GHz) and measure the rectified voltage along the b axis (Methods). The color maps in Figs. 2d and 2e reveal that the maximum rectified voltage is consistently observed throughout 110–260 GHz when the EM wave aligns with the a axis, consistent with the symmetry constraints of the NLHE. Measurements of the return loss of the 110–170 GHz frequency extender integrated with the horn antenna show increased reflection from approximately 130 GHz to 140 GHz (Supplementary Note 9), leading to reduced incident power at the device and a corresponding reduction in VDC in this frequency range. The continued response across this broad frequency range substantiates the inherent nonlinearity of quantum materials as the origin for the rectification signal, enabling operation beyond the frequency limitations imposed by junctions or barrier structures26,46. As control experiments, the same 110–260 GHz EM waves are directed at Pt and Cu devices, but no observable rectification response is detected (Supplementary Note 10). Next, we evaluate the rectified power of the TaIrTe4 device. A load resistor is connected to the TaIrTe4 device, as shown in the inset of Fig. 2f. We vary the load resistance RL and monitor the rectified power \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Pb−aaDC across the load. The maximum \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Pb−aaDC is achieved when RL matches the resistance of TaIrTe4 along the b axis, displayed in Fig. 2f. To quantitively understand the rectified power across the load, we describe \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Pb−aaDC using the following relationship27:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{{\rm{DC}}}}=\frac{{\sigma }_{b-{aa}}^{2}{E}_{a}^{4}}{4{(\frac{{\sigma }_{{bb}}}{{L}_{b}}+\frac{1}{{L}_{a}{L}_{c}{R}_{L}})}^{2}{R}_{L}}$$\end{document}Pb−aaDC=σb−aa2Ea44(σbbLb+1LaLcRL)2RLwhere \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma }_{b-{aa}}$$\end{document}σb−aa is the second-order conductivity, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\sigma }_{{bb}}$$\end{document}σbb is the first-order conductivity along the b axis, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{L}$$\end{document}RL is the resistance of the load, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{a}$$\end{document}Ea is the amplitude of incident EM field along the a axis, and La, Lb, and Lc represent the length, width and thickness of the rectifier, respectively (Supplementary Note 7). Equation (1) fits well with the RL-dependent signal in Fig. 2f and shows that \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{\sigma }_{b-{aa}}}{{\sigma }_{{bb}}}$$\end{document}σb−aaσbb is the intrinsic material parameter that determines the rectification efficiency. The resistance of many existing broadband rectifiers typically ranges from 1.2 to 7 kΩ2,8,12,14,47,48, which often requires additional impedance-matching networks to maximize efficiency49. In contrast, the resistance along the b axis of TaIrTe4 is measured to be 169 Ω, which is closer to the 377 Ω of free space. The relatively low resistance of TaIrTe4 is suitable for on-chip integration with standard RF systems. Geometric and array scaling of NLHE rectifiers To improve the rectification output, we examine how device dimensions and array configurations influence the rectified voltage. The NLHE induced rectified voltage VDC increases linearly with the device width Lb (Supplementary Note 7)2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}^{{DC}}=\frac{{\sigma }_{b-{aa}}}{{\sigma }_{{bb}}}{L}_{b}{E}_{a}^{2}$$\end{document}VDC=σb−aaσbbLbEa2 To verify this relationship, we pattern electrodes along the b-axis with increasing separation distances on the same TaIrTe4 flake, as depicted in Fig. 3a. The experimental results, shown in Fig. 3b, reveal a linear increase in the rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC with increasing the device width Lb. The result confirms that engineering Lb is an effective strategy for enhancing the output voltage.Fig. 3Impact of device width and array configurations on rectification in TaIrTe4.a Schematic illustration of the device width Lb-dependent rectification in TaIrTe4. The device width Lb is defined as the distance between the electrodes aligned along the b axis. b The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC as a function of the device width Lb of a TaIrTe4 device. c Schematic illustration of the rectification for three TaIrTe4 devices in series connection. The scale bar in the optical image is 8 μm. d \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC and rectified power \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{DC}}$$\end{document}Pb−aaDC as a function of the number of TaIrTe4 devices in series connection. The scale bar in the optical image is 4 μm. e Polarization dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC across 110-170 GHz of two TaIrTe4 devices with the a axis aligned in the same direction. f Polarization dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC across 110-170 GHz of the same two TaIrTe4 devices in (e) with a misalignment angle of 25°. The error bars in (b and d) reflect the standard deviation of five measurements. Impact of device width and array configurations on rectification in TaIrTe4. a Schematic illustration of the device width Lb-dependent rectification in TaIrTe4. The device width Lb is defined as the distance between the electrodes aligned along the b axis. b The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC as a function of the device width Lb of a TaIrTe4 device. c Schematic illustration of the rectification for three TaIrTe4 devices in series connection. The scale bar in the optical image is 8 μm. d \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC and rectified power \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{DC}}$$\end{document}Pb−aaDC as a function of the number of TaIrTe4 devices in series connection. The scale bar in the optical image is 4 μm. e Polarization dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC across 110-170 GHz of two TaIrTe4 devices with the a axis aligned in the same direction. f Polarization dependent \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{DC}}$$\end{document}Vb−aaDC across 110-170 GHz of the same two TaIrTe4 devices in (e) with a misalignment angle of 25°. The error bars in (b and d) reflect the standard deviation of five measurements. Next, we examine the effect of array configurations on rectification output enhancement. Several TaIrTe4 devices are aligned along the a axes in the same direction, and their b axes are connected in series, illustrated in Fig. 3c. The rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC is measured as a function of the number of devices in series connection. As shown in Fig. 3d, both \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Pb−aaDC increase linearly with increasing number of devices, which is consistent with expectations for series connections. We also arrange the devices in a parallel configuration and measure the output voltage and power. In this case, the output voltage remains unchanged regardless of the number of devices, while the output power scales with the number of devices (Supplementary Note 11). These findings demonstrate the scalability of NLHE rectifiers through both series and parallel configurations. In the real-world scenario, EM waves from multiple ambient sources often exhibit arbitrary polarizations, making non-polarized rectification highly desirable for energy harvesting applications. To explore this, we first align the a axes of two devices and measure the rectified voltage \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC (Fig. 3e). We then rotate one device by 25° and measure \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC again (Fig. 3f). In this rotated configuration, the a axis of the rotated device is misaligned by 25° from the incident EM wave, while the a axis of the fixed device remains aligned. After rotation, the distinct two-fold polarization dependence of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC is significantly reduced, particularly at lower frequencies. While the magnitude of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${V}_{b-{aa}}^{{{\rm{DC}}}}$$\end{document}Vb−aaDC decreases slightly due to an average 10% input power loss in the rotated device, this configuration results in a less polarization dependence. The array configurations enhance the power output and respond effectively to arbitrarily polarized EM waves, demonstrating the scalability and tunability of NLHE rectifiers. Ultra-broadband rectification and energy harvesting We investigate even higher frequency bands, extending to THz range. Using the radiation from a free-electron laser, we explore the polarization-dependent rectification signal in sub-THz to THz band using the device shown in Fig. 4a. Figure 4b shows the selected rectification response from 0.345 THz to 2.88 THz, which exhibits a clear two-fold periodicity, consistent with the symmetrical constraints in NLHE. Power-dependent rectification measurements in the THz range are conducted, which exhibit the linear dependence of the rectified voltage on the input power (Supplementary Note 12), confirming the robustness of the nonlinear Hall response in the THz frequency range. Resistance measurements under THz irradiation confirm that THz-induced temperature variation is negligible (Supplementary Note 13), and the observed rectification is dominated by the intrinsic nonlinear Hall response.Fig. 4Ultra-broadband rectification in TaIrTe4 and demonstration of energy harvesting with a thermistor.a Optical microscopy image of a TaIrTe4 device for rectification measurements. The crystallographic a and b directions of TaIrTe4 are indicated by arrows. The scale bar is 4 μm. b Polarization dependent rectified voltage at multiple sub-THz bands induced by free-electron laser. c Power conversion efficiency collected from multiple TaIrTe4 devices, with an input EM wave across 19 MHz to 2.88 THz. d Schematic illustration of energy harvesting setup with a thermistor (RTh) and a TaIrTe4 rectifier. The EM wave is emitted to a TaIrTe4 rectifier at 5.9 GHz. e The temperature-dependent resistance (RTh) of the thermistor when the heater is turned on and off alternatively shown in the lower panel. Ultra-broadband rectification in TaIrTe4 and demonstration of energy harvesting with a thermistor. a Optical microscopy image of a TaIrTe4 device for rectification measurements. The crystallographic a and b directions of TaIrTe4 are indicated by arrows. The scale bar is 4 μm. b Polarization dependent rectified voltage at multiple sub-THz bands induced by free-electron laser. c Power conversion efficiency collected from multiple TaIrTe4 devices, with an input EM wave across 19 MHz to 2.88 THz. d Schematic illustration of energy harvesting setup with a thermistor (RTh) and a TaIrTe4 rectifier. The EM wave is emitted to a TaIrTe4 rectifier at 5.9 GHz. e The temperature-dependent resistance (RTh) of the thermistor when the heater is turned on and off alternatively shown in the lower panel. To highlight the broadband rectification capabilities enabled by NLHE, we summarize the performance of TaIrTe4 devices across an ultrabroadband frequency range. In Supplementary Note 14, we present the rectified voltage, rectified power and sensitivity across 19 MHz to 2.88 THz. Figure 4c demonstrates the corresponding power conversion efficiency. Specifically, at 5.9 GHz, the TaIrTe4 rectifier exhibits a power conversion efficiency of approximately 2.6%, which is lower than that of conventional Schottky diodes (Supplementary Note 15) and CMOS rectifier (Supplementary Table 1). However, the TaIrTe4 rectifier demonstrates an exceptionally broad operational range, outperforming advanced rectification technologies such as Schottky diodes, FET diodes, ballistic rectifiers, and metal-insulator-metal diodes (Supplementary Note 16). Such a wide frequency range covers a variety of broadband applications (Supplementary Table 2). Finally, to showcase the energy harvesting capability of the NLHE rectifier, we utilize the TaIrTe4 rectifier to power a thermistor, a temperature sensor whose resistance varies with temperature. Figure 4d presents the schematic setup of the measurement. A 5.9 GHz EM wave is directed onto the rectifier, which is connected with the thermistor as a power source. The current meter and the voltmeter used to monitor the thermistor's current and voltage are passive elements in this circuit and do not inject power into the thermistor or the rectifier. The temperature of the thermistor alternates between 20 and 60 °C using a controlled heater. Throughout the experiment, the resistance RTh of the thermistor changes in response to temperature fluctuations, as shown in Fig. 4e. The results confirm the proper operation of the thermistor powered solely by the TaIrTe4 rectifier, demonstrating the energy-harvesting capability of a nonlinear Hall rectifier. Broadband frequency mixing based on NLHE Mixers are essential components in wireless communication systems, which have been widely used for wireless networks, radar, and satellite communications50,51. A frequency mixer combines an RF signal with a local oscillator (LO) signal to generate sum and difference frequency components through nonlinear interactions. The schematic of the NLHE mixer is shown in Fig. 5a. The device is designed with two pairs of electrodes along the a axis of TaIrTe4 for injecting RF and LO signals, while the output is measured along the b axis, corresponding to the transverse nonlinear Hall component (Methods). We select 12.5 GHz as the LO signal and 18 GHz as the RF signal, both within the Ku band, commonly used in satellite communications and radar applications. Figure 5b presents the output power of the tuned RF signal and conversion gain (CG) as a function of the input power of LO, with the input power of RF fixed at 0 dBm. The NLHE mixer exhibits a CG of around –20 dB at the up-converted frequency of 30.5 GHz and around –23dB at the down-converted frequency of 5.5 GHz. In addition, the measured input 1-dB compression point (IP1dB) of 4 dBm indicates relatively high linearity, which is crucial for minimizing intermodulation distortion in high-frequency signal processing. To further assess the broadband performance of the device, we measure CG across a wide range of LO frequencies with a fixed RF frequency of 18 GHz, as shown in Fig. 5c. The mixer demonstrates up- and down-conversion over a broad LO range from 0.1 GHz to 40 GHz. The broadband nature of the response is a direct consequence of the intrinsic material property, which is not limited by thermal voltage threshold or the transition time of the semiconductor junction26. We note that the first demonstration of NLHE passive mixers was reported in NbP and Pt22, focusing on wired signal mixing at low frequencies without reporting key parameters such as CG or linearity (Supplementary Table 3), and wireless frequency mixing shown in Td-NbIrTe4 device36. Our work builds on this concept by achieving broadband frequency mixing with defined performance metrics. The CG and broadband operation of our device are comparable to those reported in passive mixer devices and serve as a useful benchmark for NLHE mixer performance (Supplementary Table 4).Fig. 5Frequency mixing driven by NLHE in TaIrTe4.a Schematic of the wired frequency mixing setup. Two input signals representing the local oscillator (LO, ωLO) and radio frequency (RF, ωRF) are injected into the electrodes along the a axis of TaIrTe4, and the output power is measured along the b axis of TaIrTe4. b Output power (Poutput) and conversion gain (CG) as a function of the input power of LO. The dashed line at 4 dBm indicates the input power 1 dB compression point at 4 dBm. c CG as a function of input frequency of LO from 0.1 to 40 GHz, showing both up-converted and down-converted frequencies. The dashed line at –20.6 dB represents the average value of CG for the down-converted frequencies. d Output power of the wireless frequency mixing signal measured by the spectrum analyser (SA). Peaks marked with black asterisks originate from wireless LAN signals in the ambient environment. The inset illustrates the wireless frequency mixing setup. Frequency mixing driven by NLHE in TaIrTe4. a Schematic of the wired frequency mixing setup. Two input signals representing the local oscillator (LO, ωLO) and radio frequency (RF, ωRF) are injected into the electrodes along the a axis of TaIrTe4, and the output power is measured along the b axis of TaIrTe4. b Output power (Poutput) and conversion gain (CG) as a function of the input power of LO. The dashed line at 4 dBm indicates the input power 1 dB compression point at 4 dBm. c CG as a function of input frequency of LO from 0.1 to 40 GHz, showing both up-converted and down-converted frequencies. The dashed line at –20.6 dB represents the average value of CG for the down-converted frequencies. d Output power of the wireless frequency mixing signal measured by the spectrum analyser (SA). Peaks marked with black asterisks originate from wireless LAN signals in the ambient environment. The inset illustrates the wireless frequency mixing setup. Beyond conventional wired signal injections, we explore the wireless frequency mixing capabilities of the device. RF and LO signals are transmitted via antennas, as shown in the inset of Fig. 5d (Methods). We select 12.5 GHz as the LO signal and 18.8 GHz as the RF signal. As shown in Fig. 5d, the up-converted (31.3GHz) and down-converted (6.3GHz) frequencies can be clearly observed, confirming the feasibility of the NLHE-based mixer for wireless applications. The power-dependent mixing signal is consistent with a nonlinear Hall origin (Supplementary Note 17), and the linear I-V characteristics of the device exclude contributions from Schottky junctions (Supplementary Note 18). The peaks at lower frequency of 925 MHz, 2.45 GHz, and 5.3 GHz, come from wireless LAN in free space due to the rectification capability52. We have demonstrated broadband rectification and frequency mixing based on NLHE in TaIrTe4. Specifically, wireless rectification across an ultrabroadband spanning from 19 MHz to 2.88 THz at room temperature with zero bias voltage and zero magnetic field is obtained. Utilizing a series connection, we have enhanced the rectified voltage and reduced the sensitivity of rectifiers to the polarization. In addition, we have realized NLHE-based passive mixer with broadband, high linearity, and zero DC bias. The NLHE rectifiers and mixers offer advantages not only in an extensive frequency range but also in a simple single-layer structure, which eliminates the need for creating Schottky barriers or tunnelling junctions. Looking ahead, based on the scalability of the NLHE rectifier array, we anticipate the development of NLHE rectifiers made from thin films, capable of supporting wafer-scale array topologies. Such advancements would require thin-film materials with high second-order nonlinear Hall conductivity and low first-order conductivity along the Hall direction (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{{\sigma }_{b-{aa}}}{{\sigma }_{{bb}}}$$\end{document}σb−aaσbb) at room temperature. Moreover, rectification performance can be further optimized through a series of engineering strategies, such as integrating optimized antennas, designing impedance-matching circuits, and incorporating low-pass filters to maximize energy harvesting efficiency. This work establishes a strong foundation for the practical application of NLHE in ultrabroadband rectification and frequency mixing, paving the way for advancements in next-generation energy harvesting systems, wireless sensing, and low-power communication systems. Methods Growth of TaIrTe4 TaIrTe4 single crystals were synthesized via a self-flux method using elemental Ta powder (99.99%), Ir powder (99.999%), and Te lump (99.999%) in an atomic ratio of 1:1:12. The stoichiometric mixture was loaded in a quartz tube and then flame-sealed under a high vacuum of 10−6 Torr. Then, the quartz tube was heated to 1,323 K and held for 7 days, and slowly cooled to room temperature for 15 days to ensure a complete reaction. Subsequently, the samples were transferred to another evacuated quartz tube with quartz wool under a vacuum of 10-6 Torr, which was centrifuged at 873 K to divide the TaIrTe4 crystals from the surroundings of excess Te. Finally, bar-shaped TaIrTe4 single crystals with metallic lustre were obtained. Raman spectroscopy The Raman spectra were acquired using a Renishaw inVia Raman microscope. A linearly polarized 532 nm excitation laser was focused onto the samples through a 50× objective lens. The scattered Raman signals were collected through the same lens and directed back to the spectrometer in a parallel configuration. The sample was mounted on a rotation stage and rotated with a step size of 10°. All parameters were kept identical throughout the measurements to ensure consistency. Device fabrication TaIrTe4 samples were prepared by mechanically exfoliating single-crystalline bulk Td-TaIrTe4 onto Si substrates (1–10 Ω·cm) with a 300 nm thermally grown SiO2 layer. Rectangular-shaped flakes were selected for investigation. A 5 nm thick SiO2 capping layer was deposited on the flakes using sputtering to prevent material oxidation and degradation, with a base pressure of less than 2 × 10-9 Torr. The flake thickness was characterized by atomic force microscopy. For electrical transport measurements, standard Hall bar geometries were patterned using electron-beam lithography (EBL) and etched using argon ion milling. Electrical contacts with a Ta (5 nm)/Cu (70 nm)/Ta (5 nm) structure was patterned by EBL and an ultraviolet maskless lithography machine (TuoTuo Technology, UV Litho-ACA), and deposited via sputtering. For devices for rectification and frequency-mixing measurements, electrical contacts were directly patterned and deposited onto exfoliated flakes without prior etching into Hall-bar geometries. The same contact structure and fabrication tools were utilized for both transport and rectification devices, ensuring consistency throughout the device fabrication process. Electrical transport measurements The electrical transport measurements were conducted at room temperature under zero magnetic field. During the measurement, a constant-amplitude sinusoidal current with a frequency of 13.7 Hz was applied to the devices (Keithley 6221). Fundamental and second-harmonic frequencies of both longitudinal and transverse voltages were simultaneously recorded by four lock-in amplifiers (Stanford Research SR830). Rectification measurements For rectification at 19 to 40 GHz, the EM wave was emitted from a signal generator (Agilent E8257D) and transmitted to a broadband horn antenna (Enrichtek ATBDH10180 for 1-18 GHz, and ATBDH18400 for 18-40 GHz). The EM wave below 1 GHz was delivered through commercial broadband antennas (MKZBH). For rectification at 110-260 GHz, the RF signal was irradiated from a vector network analyzer (Rohde & Schwarz ZNA26) with an extender and delivered to broadband horn antennas (Eravant SAR-2507-06-S2 for 110-170 GHz, and SAR-2507-04-S2 for 170-260 GHz). The device was placed parallel to and at the center of the emission surface of the pyramidal antenna to receive the linearly polarized radiation with normal incidence. The rectified voltage was recorded by a nanovoltmeter (Keithley 2182 A) at room temperature with zero magnetic field and zero electrical bias. To ensure an unbiased assessment, the input power Pin is defined as the total power output from the EM source (signal generator, vector network analyzer, or free-electron laser source). This definition does not account for losses at the device interface, the micro-meter scale device area receiving the power, power attenuation due to air, or the gain of the transmitting antenna. In polarisation-dependent measurements, the device was rotated by a motorised precision rotation stage (Thorlabs PRM1Z8). The direction of the EM wave was fixed along the a-axis of TaIrTe4 unless otherwise stated. For power-dependent measurements, the input power Pin was gradually increased from the signal generator. The input power Pin was fixed to 5 mW unless otherwise specified. For THz-range measurements, the source is the free-electron laser (FLARE) located at Radboud University. Its radiation comes in the form of 10 μs long macropulses consisting of picosecond-scale micropulses at 3 GHz repetition rate. Linearly polarized light from the laser was focused into a 5 mm diameter focal beam spot and irradiated the device at normal incidence. The voltage signal was recorded by the digital oscilloscope (PXIe-5162) at room temperature with zero magnetic field and zero electrical bias. Similar to RF rectification, the input power Pin is defined as the actual power output from the free electron laser source. The polarisation-dependent measurements used the same motorised precision rotation stage to rotate the device. For power-dependent measurements, the emitted power was adjusted by an attenuator. The rectified voltage was measured between two contacts of the TaIrTe4 device using a Keithley 2182 A nanovoltmeter. The HI (High) and LO (Low) inputs of the nanovoltmeter are connected to the two electrodes of the device, labeled as V+ and V- in the optical images, respectively. The LO input of the nanovoltmeter is internally connected to the instrument ground, and the instrument is connected to earth ground via the rear-panel analog output. As a result, the TaIrTe4 device electrode connected to the LO input (V-) is grounded through the nanovoltmeter, while the device itself does not include a separate ground pad. Frequency mixing measurements The input RF and LO signals were emitted from two signal generators (Agilent E8257D). To ensure an unbiased assessment, the input power is defined as the actual power output from the EM source. For conventional wired measurements, the EM wave was directly injected into the device through bonded wires. For wireless measurements, the EM wave was delivered to the device through two broadband antennas (Enrichtek ATBDH10180 and ATBDH18400). The output power is measured by a spectrum analyzer (Agilent E4448A). Supplementary information Supplementary Information Transparent Peer Review file Supplementary Information Transparent Peer Review file