Symmetry-cooperative optical-field engineering for ultrabroadband bulk photovoltaic photodetection

Abstract

The bulk photovoltaic effect (BPVE) enables light-to-electricity conversion without external bias, but its practical reach is constrained by the rapid collapse of photoresponse at low photon energies. Here we show that this limitation can be overcome through cross-scale symmetry cooperation between chiral tellurium nanowires and an asymmetric plasmonic microstructure. The nanowire supplies a crystal-allowed photovoltaic current channel, while the microstructure redistributes optical excitation into a spatially asymmetric near field aligned with that channel, amplifying the measurable zero-bias response without altering the intrinsic material mechanism. The resulting devices operate at room temperature across an ultrabroad spectral range, from the visible to 0.22 THz, and retain a clear terahertz response far below the interband regime. Near-field imaging, polarization-resolved measurements, electromagnetic simulations and symmetric controls consistently support the mechanism. These results establish device-scale optical-field symmetry as a powerful design degree of freedom for extending bulk photovoltaic photodetection into low-energy spectral regimes.

The authors establish device‑scale optical‑field symmetry asa powerful design degree of freedom for extending bulk photovoltaic photodetection into low‑energy spectral regimes atroom temperature.

Introduction

The bulk photovoltaic effect provides a direct route for bias-free conversion of optical excitation into dc electrical signals in non-centrosymmetric materials, offering distinct advantages for self-powered photodetection and energy conversion^1,2. Unlike conventional photovoltaic mechanisms that rely on built-in junctions or external bias, BPVE arises from symmetry-allowed even-order photoresponses and is intrinsically linked to the crystallographic properties of the active medium^3–6. Microscopically, BPVE is governed by quantum geometric properties in momentum space⁷. In particular, photoexcitation can induce a gauge-invariant real-space displacement of the electronic wavepacket described by the shift vector, a gauge-invariant quantity rooted in the Berry connection and interband optical geometry⁸. The resulting macroscopic current can be quantified by the bulk photovoltaic coefficient β, reflecting the strength of the underlying quantum-geometric response⁹. Over the past decade, extensive efforts have successfully identified and engineered a wide range of materials with enhanced intrinsic BPVE responses, including ferroelectrics^10,11, polar semiconductors¹², topological systems^13–15, and chiral crystals^16,17, leading to substantial advances in the visible and infrared spectral regimes.

Building on this material-centered progress, recent studies have also explored various approaches to further optimize BPVE responses, such as band-structure engineering¹⁸, strain^9,19, electrostatic gating²⁰, and moiré superlattices²¹. These strategies have demonstrated that the magnitude and directionality of bulk photovoltaic currents can be sensitively tuned through modifications of electronic structure and symmetry. Nevertheless, in most reported systems, efficient BPVE operation remains closely associated with interband optical transitions, which naturally constrains device performance as the photon energy is reduced^4,6,22. Consequently, extending BPVE-based photoresponse into low-energy regimes, such as the far-infrared and terahertz (THz) bands continues to be challenging, despite the impressive material advances achieved to date.

In parallel with advances in materials, optical excitation in practical devices is mediated by spatially structured electromagnetic fields that can be strongly reshaped by micro- and nanostructured environments^23,24. While such optical-field engineering has been widely employed to enhance absorption, localization, and light–matter interaction in photonic and optoelectronic devices, its potential role in modulating bulk photovoltaic responses has received comparatively less attention. In particular, a timely question is whether device-scale redistribution of optical excitation can be harnessed to cooperate with material symmetry. This cooperation would be achieved by directing energy deposition and carrier displacement in a symmetry-compatible manner, a strategy particularly needed in low-energy regimes where conventional interband-driven mechanisms are less effective.

Here, we address this question by integrating chiral tellurium (Te) nanowires with asymmetric microstructured environments designed to produce controlled optical-field asymmetry. Tellurium is a representative elemental chiral semiconductor with helically arranged atomic chains, non-centrosymmetric enantiomorphic crystal structures, and a well-defined chiral axis^22,25. We cite previously reported BPVE-type responses in low-dimensional Te-based systems, including tellurene, as supporting evidence for Te-based non-centrosymmetric lattices, rather than as a direct demonstration of the same nanowire device system. By coordinating the intrinsic crystal symmetry of Te with structure-induced spatial redistribution of optical excitation, we show that device-scale optical-field engineering offers an additional degree of freedom to increase the measurable zero-bias dc photoresponse, without modifying the intrinsic crystal-level BPVE mechanism. Our room-temperature detector exhibits measurable zero-bias photoresponse spanning discrete bands from the visible to 0.22 THz, with device-level effective BPV coefficients of 11.7 V⁻¹ at 638 nm, 0.70 V⁻¹ at 1550 nm and 0.44 V⁻¹ at 0.033 THz. These results provide a practical design perspective for extending BPVE-enabled photoresponse towards low photon energies by synergistically combining crystal asymmetry with engineered optical-field redistribution, complementing material-centric approaches for symmetry-enabled, bias-free photodetection.

Results

Synergistic asymmetry: matching intrinsic Te symmetry with device-scale optical-field redistribution

To establish a platform in which intrinsic bulk photovoltaic response can be systematically reinforced at the device level, we design a synergistically asymmetric structure device (SASD) that integrates a non-centrosymmetric semiconductor with a deliberately asymmetric electromagnetic environment. Here, synergistic asymmetry is used as a descriptive device-design concept rather than as a new intrinsic BPVE mechanism. It refers to the cooperation between two sources of asymmetry: the non-centrosymmetric chiral lattice of Te, which defines a symmetry-allowed zero-bias BPVE/photogalvanic current channel, and the asymmetric Au microstructure, which redistributes the incident optical field at the device scale. Cross-scale symmetry cooperation further denotes the matching between the crystal-scale current axis and the device-scale optical-field asymmetry, so that local optical excitation can be more efficiently projected onto the two-terminal collection direction. As schematically shown in Fig. 1a, the SASD consists of a Te nanowire transferred onto pre-patterned Au disk microstructures and strip electrodes. The nanowire directly contacts the Au structures at the overlap regions, whereas the channel region can partially bridge above the substrate because of the finite metal thickness. This configuration is chosen to coordinate the crystallographic asymmetry of the active material with structure-induced optical-field redistribution, rather than relying on either contribution alone.

Fig. 1. Synergistic asymmetry via coupled material and structural polarization.

a Schematic of the synergistically asymmetric structure device (SASD), consisting of a Te nanowire transferred onto pre-patterned Au disk microstructures and strip electrodes. The device is designed to coordinate intrinsic crystal asymmetry with structure-induced optical-field redistribution. The nanowire is electrically contacted by the Au structures, while the channel region can partially bridge above the substrate. b Atomic models of right- and left-handed helical Te chains. The corresponding interlayer differential charge density (red, accumulation; green, depletion) highlights the spontaneous polarization along the crystallographic c-axis arising from broken inversion symmetry. c Finite-difference time-domain (FDTD) simulation of the optical near-field intensity under uniform illumination, showing localized field enhancement at the edges of the metallic microstructures. d Simulated differential charge density in the active region of the SASD, illustrating the cooperative contribution of intrinsic material polarization and structure-induced optical-field asymmetry to axial charge separation. e Polarization-resolved near-field intensity of the asymmetric microstructure and second-harmonic generation (SHG) response of the Te nanowire under 1064 nm excitation. The consistent angular dependence reflects the non-centrosymmetric character of the coupled system.

Tellurium is a representative elemental chiral semiconductor whose helical atomic chains lack inversion symmetry, giving rise to a spontaneous polarization along the crystallographic c-axis. First-principles calculations of the interlayer differential charge density reveal a pronounced charge redistribution between adjacent atomic layers (Fig. 1b), directly visualizing the microscopic origin of this intrinsic polarization. Consistent with this picture, the spatial distributions of the valence- and conduction-band wavefunctions exhibit marked anisotropy along the chain direction (Supplementary Fig. S1), confirming that the optical response of Te is inherently symmetry selective. The atomic model is shown in the same crystallographic orientation as the differential charge-density map. Supplementary Fig. S6 provides the corresponding helical-chain structural reference, and Supplementary Fig. S1 shows the related band-edge charge-density distributions.

Beyond the material itself, optical excitation in the SASD is shaped by the asymmetric metallic microstructures. The disk-strip geometry intentionally breaks in-plane inversion symmetry, leading to a polarized and spatially inhomogeneous near-field under illumination. FDTD simulations show that the optical field is strongly localized at the edges of the metallic elements (Fig. 1c and Supplementary Fig. S2), establishing controlled field gradients in the vicinity of the nanowire. Such structure-induced redistribution of optical excitation provides an additional handle for directing energy deposition at the device scale.

The cooperative effect of material and structural asymmetry is illustrated in Fig. 1d, which shows the simulated differential charge density in the active region of the SASD. When the intrinsic polarization of the Te nanowire is aligned with the optically induced asymmetry imposed by the microstructure, an enhanced axial charge separation emerges, reflecting the synergistic contribution of the two symmetry-breaking elements. Importantly, this enhancement arises from their coordinated action, rather than from either material or structure alone.

We further verify the symmetry properties of the coupled system using polarization-resolved optical probes. SHG, a well-established signature of broken inversion symmetry⁸, exhibits a characteristic angular dependence for the Te nanowire under 1064 nm excitation (Fig. 1e). Notably, the polarization-dependent near-field intensity of the asymmetric microstructure follows a consistent angular trend, indicating that the structure-induced optical-field asymmetry is symmetry-compatible with the intrinsic response of the material. Together, these observations confirm that the SASD realizes a coherently symmetry-broken system, providing a well-defined foundation for the enhanced zero-bias photoresponse explored in the following sections.

Microstructure-integrated Te nanowire devices with symmetry-broken near-field confinement

To experimentally examine how device-scale optical-field engineering influences the bulk photovoltaic response, we first establish a well-defined material and device platform based on high-quality Te nanowires. Te nanowires were synthesized by chemical vapor deposition using TeO₂ as the precursor (Supplementary Fig. S3), yielding uniform and crystalline structures with lengths exceeding hundreds of micrometers. Individual nanowires were subsequently transferred and aligned onto pre-patterned substrates using a dry-transfer process (Supplementary Fig. S4). The Au disks and strip electrodes are fabricated in the same lithography and metallization process and therefore have nearly identical heights. The Te nanowire directly contacts the Au structures at overlap regions, while the channel can partially bridge above the substrate (Supplementary Fig. S5).

Figure 2a shows a representative Te nanowire with a width of approximately 8 μm, together with an optical image of the fabricated SASD, in which the nanowire is integrated with engineered Au microstructures. The intrinsic structural asymmetry of Te is a prerequisite for bulk photovoltaic response. As illustrated by the atomic models (Supplementary Fig. S6), Te crystallizes in a helical chain configuration within a hexagonal lattice, resulting in broken inversion symmetry along the chain (c) axis. This chiral atomic arrangement is directly resolved by high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), which reveals the characteristic helical ordering at the atomic scale (Fig. 2b and Supplementary Fig. S7). Complementary polarization-dependent Raman spectroscopy further confirms the anisotropic vibrational response associated with the quasi-one-dimensional helical chains (Fig. 2c), providing optical evidence of the crystallographic anisotropy and non-centrosymmetric nature of the nanowires.

Fig. 2. Microstructure-integrated Te nanowire devices with symmetry-broken near-field confinement.

a Top: high-resolution SEM image of a Te nanowire with a width of ~8 μm. Bottom: optical micrograph of a representative synergistically asymmetric structure device (SASD), showing the integration of the Te nanowire with patterned Au microstructures. b Medium- and high-resolution HAADF-STEM images revealing the chiral helical atomic arrangement of the Te nanowire. c Polarization-resolved Raman intensity mapping as a function of incident angle, confirming the crystalline anisotropy and broken inversion symmetry of the Te nanostructure. d Schematic illustration of the Au microstructure geometry, consisting of a periodic disk array (thickness 100 nm, period 1.5 μm) connected to a strip electrode and integrated with the transferred Te nanowire. e Top: SEM image of the fabricated Au nanodisk array. Bottom: corresponding height profile along the dashed line, confirming the structural morphology. f FDTD simulation of the optical near-field intensity under 1200 nm, x-polarized illumination, showing localized field enhancement near the microstructure edge. g Line profile of the simulated optical field intensity along the dashed line in (f), highlighting the directional asymmetry of the near-field distribution. h Top: schematics of the SSD and the SASD. Bottom: I–V curves of the SSD and SASD under 1550 nm flood illumination (power density = 3.66 W/cm²). i, j Spatial photocurrent mapping and corresponding line profiles (along the white dotted lines) for the SSD and SASD under 1550 nm local illumination (power density =  15.98 W/cm²).

While the intrinsic lattice asymmetry enables a shift current²², efficient BPVE activation—especially in the infrared regime—demands strong local field confinement and additional symmetry breaking at the device scale. To introduce a controlled source of asymmetry at the device level, we design and integrate plasmonic Au microstructures adjacent to the transferred Te nanowire. As schematically shown in Fig. 2d, the microstructure consists of a periodic array of Au nanodisks (thickness 100 nm, period 1.5 μm) positioned near one terminal of the nanowire and connected to a strip electrode. Scanning electron microscopy and atomic force microscopy measurements confirm the uniformity and well-defined geometry of the nanodisk array (Fig. 2e). This asymmetric configuration is intentionally chosen to reshape the local electromagnetic environment experienced by the nanowire under optical excitation. Although isolated microstructure units are shown for geometric clarity, the actual SASD uses a connected disk-strip geometry. This interconnection modifies the electromagnetic boundary conditions and breaks the lateral symmetry of the metallic environment, producing a more directional and spatially asymmetric near-field distribution near the microstructure-strip interface.

The impact of the microstructure on the optical near-field distribution is examined by FDTD simulations. Under x-polarized illumination at 1200 nm, the optical field exhibits pronounced localization near the edges of the Au nanodisks and at their junction with the strip electrode (Fig. 2f). The normalized field amplitude |E/E₀| and field intensity |E/E₀|² are explicitly distinguished. The corresponding line profile (Fig. 2g) reveals a clear directional asymmetry in the field intensity, indicating that the microstructure induces a spatially polarized near-field aligned along the nanowire axis. Under 1200 nm illumination, the maximum field-amplitude enhancement along the polarization direction reaches approximately 3, corresponding to a field-intensity enhancement of approximately 9. This enhancement should not be interpreted only as peak absorption enhancement; the asymmetric microstructure also redistributes the spatial excitation weight and creates a directional field gradient along the Te nanowire axis. This redistribution of optical excitation provides a direct means of breaking spatial symmetry at the device scale, beyond the intrinsic crystal asymmetry of the material. FDTD-simulated near-field intensity at different wavelengths see Supplementary Fig. S8a–c. We also compared representative alternative microstructure geometries, including smaller-diameter disk arrays and rectangular arrays. The adopted asymmetric disk-strip geometry provides a favorable compromise between spatial field modulation, broadband stability, and fabrication compatibility (Supplementary Fig. S8d–i).

To directly assess the influence of this engineered asymmetry on device response, we fabricate and compare two architectures: an SSD featuring dual-strip electrodes, and the SASD incorporating the plasmonic microstructure (schematics in Fig. 2h, top). Under 1550 nm illumination, the SSD exhibits a weak and nearly symmetric current–voltage response, whereas the SASD shows a markedly enhanced zero-bias photocurrent and open-circuit voltage (Fig. 2h, bottom). Spatially resolved photocurrent mapping further highlights this contrast. While the SSD displays a relatively uniform and low-amplitude photocurrent distribution (Fig. 2i), the SASD exhibits a strongly asymmetric photocurrent profile with a pronounced maximum localized near the microstructure region (Fig. 2j). This spatial correlation directly visualizes the cooperative effect of intrinsic material asymmetry and structure-induced optical-field confinement, establishing the experimental basis for the enhanced photoresponse discussed in the following sections. SPCM was performed to examine how channel length and contact geometry influence the spatial distribution of the photocurrent in SSD under zero applied bias in Supplementary Fig. S9. Long-channel SSD controls in Supplementary Fig. S10 show that channel length changes the absolute collection condition but does not generate the strongly asymmetric photocurrent distribution observed in the SASD.

Furthermore, we fabricated orientation-dependent devices to test whether the enhancement depends on the matching between the Te crystal-allowed current axis and the microstructure-induced optical-field asymmetry. The aligned configuration yields the strongest zero-bias response, whereas the 60° misaligned device retains a reduced but still appreciable response, and the 180° reversed field-projection device shows a strongly suppressed photocurrent (Supplementary Fig. S11). This orientation dependence confirms that the enhancement relies critically on the relative matching between the Te current axis and the engineered optical-field asymmetry.

To distinguish the unidirectional channel response from contact- or thermal-gradient-related contributions, we performed a phenomenological decomposition of the spatial photocurrent distribution. For two spatially symmetric positions P1 and P2, the unidirectional axial component and reverse-signed component are defined as

$$I_{\mathrm{uni}}=\left(I_{\mathrm{P}1}+I_{\mathrm{P}2}\right)/2$$ 1

$$I_{\mathrm{rev}}=\left(I_{\mathrm{P}1}-I_{\mathrm{P}2}\right)/2$$ 2

respectively. Here, I_uni represents the photocurrent component with the same polarity along the Te channel, I_rev represents the opposite-polarity component expected from contact-related or thermal-gradient-induced contributions, I_P1 and I_P2 represent the currents generated at two symmetrical positions within the channel when light is incident on them. The unidirectional component dominates the SASD channel response, whereas the reverse-signed component is comparatively weak, indicating that contact- or thermal-gradient-related effects may exist as background contributions but are unlikely to dominate the enhanced zero-bias response (Supplementary Fig. S12).

To quantitatively assess the impact of synergistic symmetry breaking on device performance, we systematically compare the photoresponse of SSD and SASD detectors as a function of excitation wavelength, polarization, and optical power density. Importantly, both devices are fabricated from identical Te nanowires, ensuring that the observed differences arise from device architecture rather than material variability.

The spectral dependence of the enhancement is first examined. FDTD simulations predict wavelength-dependent near-field amplification induced by the microstructure, with prominent features around 580 and 1155 nm (Fig. 3a). Experimentally, the photocurrent enhancement of the SASD follows the same trend (Fig. 3b), reaching pronounced maxima at the corresponding wavelengths relative to the SSD. Beyond the near-infrared, the enhancement persists into the mid-infrared regime, where the SASD exhibits a substantially broadened and strengthened photoresponse compared with the SSD (Fig. 3c). We quantify the wavelength-dependent enhancement by G(λ) = |I_SASD(λ)|/|I_SSD(λ)|. G(λ) remains larger than unity over the measured visible/NIR and MIR wavelengths, reaching approximately 3 at 580 nm, 16 at 1150 nm and about 4 in the measured MIR range.

Fig. 3. Experimental verification of synergistic-asymmetry-enhanced bulk photovoltaic response.

a FDTD-simulated wavelength-dependent optical field intensity ratio E²_SASD/E²_SSD, where E²_SASD is sampled from the SASD geometry with the Au disk microstructure, and E²_SSD is sampled from the same channel region in an otherwise identical SSD control geometry without the Au disk microstructure. b Experimentally measured wavelength-dependent photocurrent enhancement of the SASD relative to the SSD, in good agreement with the simulated trend in (a). c Mid-infrared wavelength-dependent photocurrent comparison between the SASD and SSD, demonstrating enhanced sensitivity enabled by the microstructure. d Polarization-dependent short-circuit current (I_sc) of both devices at 1550 nm, highlighting the anisotropic response of the SASD arising from symmetry breaking. e Comparison of photocurrent intensity under broadband flood illumination and localized illumination, illustrating the enhancement associated with the microstructure in the SASD. f Laser power-density dependence of the photocurrent for the SASD and SSD. Solid lines indicate linear and sublinear (square-root) scaling regimes.

These results demonstrate that engineered optical-field asymmetry remains effective across a wide spectral range, including photon energies approaching the low-energy limit of the material (Supplementary Fig. S13).

Polarization-dependent measurements provide further insight into the symmetry-governed nature of the response. At 1550 nm, the short-circuit current of both devices exhibits a clear angular dependence on the incident polarization (Fig. 3d). While the SSD shows a relatively weak modulation, the SASD displays a markedly enhanced anisotropic response, consistent with the introduction of an additional symmetry-breaking element at the device level.

To distinguish the role of the microstructure from trivial geometric or shadowing effects, we compare device responses under broadband flood illumination and tightly focused local illumination. Even when the excitation is confined to a small spot that minimizes direct interaction with the microstructure, the SASD consistently produces larger photocurrents than the SSD over a broad spectral range (Fig. 3e). This observation indicates that the enhancement is not solely attributable to increased illuminated area, but reflects a genuine modification of the photoresponse enabled by the engineered electromagnetic environment.

The nonlinear nature of the photoresponse is further examined through power-dependent measurements. As shown in Fig. 3f and Supplementary Fig. S14, both devices exhibit an approximately linear dependence at low power and a sublinear evolution at higher power. Such behavior is characteristic of even-order rectified photoresponses in non-centrosymmetric systems and is consistent with bulk photovoltaic-type mechanisms reported in related materials^5,6,8,19. But it is not used here as a standalone criterion for assigning the mechanism, because photothermoelectric or other thermally related effects can also show similar power dependences. Therefore, the power-dependent data are considered together with SSD/SASD control devices, spatial photocurrent mapping, temperature-dependent measurements, optical-field simulations, and orientation-dependent devices. To quantify the response magnitude, the short-circuit current density ($J_{sc}$) can be expressed, for a fixed polarization configuration, in the form $J_{sc}=\left(A+B\right)I$, where A represents the material-intrinsic contribution, and B captures the additional contribution introduced by the microstructure. Compared with the SSD, the SASD exhibits a substantially larger effective coefficient, highlighting the role of synergistic asymmetry in amplifying the zero-bias photoresponse.

To examine whether the enhanced response is governed primarily by a thermally activated process, temperature-dependent power measurements were performed from 80 to 210 K. The photocurrent–power characteristics show no qualitative change over this temperature range. This observation is not used as a standalone proof of BPVE, but supports that a strongly temperature-activated PTE or bolometric mechanism is not the dominant origin of the enhanced zero-bias response (Supplementary Fig. S15).

We distinguish spectral accessibility from device-level enhancement. The visible-to-infrared spectral accessibility is primarily provided by the narrow-bandgap electronic structure and broadband absorption of Te. The asymmetric microstructure does not create this response from an inactive material; instead, it amplifies the measurable zero-bias output by redistributing the optical field and improving the projection of local photocurrent onto the two-terminal collection direction.

Room-temperature, zero-bias terahertz photoresponse activated by cross-scale asymmetry

We next examine the photoresponse of the SASD in the low-photon-energy regime, focusing on THz excitation well below the bandgap of tellurium. In this regime, conventional interband-driven photoresponses are strongly suppressed, and symmetric Te nanowire devices exhibit negligible signals under uniform THz illumination. This provides a stringent test for evaluating whether device-scale optical-field engineering can enable additional photoresponse pathways beyond material-intrinsic limitations. To probe the electromagnetic environment under THz excitation, we perform both numerical simulations and near-field experimental mapping at 0.127 THz. Simulations reveal a pronounced and polarization-dependent confinement of the THz field near the central region of the microstructure (Fig. 4a), indicating that the asymmetric geometry reshapes the local electromagnetic field distribution at long wavelengths. Consistently, the experimentally measured THz near-field reflection map shows strong localized response at the interface between the microstructure and the strip electrode (Fig. 4b), closely matching the simulated field profile. In contrast, symmetric devices display a largely uniform and polarization-independent near-field response (Supplementary Fig. S16), confirming that the observed field localization arises from engineered structural asymmetry.

Fig. 4. Room-temperature, zero-bias terahertz photoresponse enabled by cross-scale asymmetry.

a Simulated near-field intensity distribution under 0.127 THz illumination, showing polarization-dependent field confinement near the central microstructure. Inset: cross-sectional field profile along the indicated line. b Experimental map of the THz near-field reflection at 0.127 THz, revealing localized response at the interface between the microstructure and the strip electrode. c Time-resolved photocurrent signals of the SASD measured at selected THz frequencies (0.033, 0.10, and 0.22 THz) under broad-area illumination and zero external bias. d Frequency dependence of the short-circuit photocurrent of the SASD from 0.02 to 0.10 THz, demonstrating broadband response at room temperature. Inset: schematic illustration of intraband carrier absorption under THz excitation. e Current–voltage (I–V) characteristics of the SASD measured in the dark and under modulated 0.10 THz illumination, showing a voltage-independent photocurrent plateau.

Under these conditions, the SASD exhibits clear time-resolved photocurrent signals at multiple THz frequencies without external bias and at room temperature (Fig. 4c). The device maintains a measurable short-circuit photocurrent across a broad frequency range from 0.02 to 0.10 THz, with a maximum current exceeding 20 nA (Fig. 4d). These observations demonstrate that cross-scale asymmetry enables a robust zero-bias photoresponse in a spectral regime where conventional Te nanowire devices remain inactive. Because the SSD signal is close to the noise floor in the THz range, a finite enhancement ratio is not assigned. The THz response is therefore described as a microstructure-activated zero-bias response channel rather than as a simple amplification of the SSD response. The zero-bias terahertz photoresponse speed of SASD provides support for the assessment of light response (Supplementary Fig. S17). Current–voltage characteristics measured under modulated 0.10 THz illumination further support this interpretation. As shown in Fig. 4e and Supplementary Fig. S18, the photocurrent exhibits a nearly voltage-independent plateau over the measured bias range, distinct from photoconductive behavior that would scale with applied voltage. Together with the absence of measurable response in symmetric control devices, this behavior indicates that the THz photocurrent originates from a symmetry-enabled rectification process rather than from trivial conduction or contact effects.

From a physical perspective, the visible/infrared and THz responses should be distinguished. In the visible and near-infrared regimes, the photon energy can drive interband transitions in Te, and the zero-bias response is mainly discussed in terms of a symmetry-allowed BPVE/photogalvanic-type contribution that is enhanced at the device level by local-field redistribution. In the THz regime, the photon energy is far below the interband transition energy of Te, so conventional interband shift current is strongly suppressed. The THz field primarily drives intraband carrier motion. In the SASD, the asymmetric microstructure generates a spatially nonuniform THz near field and a directional field gradient, enabling field-gradient-assisted intraband rectification to survive device-level spatial integration and produce a measurable dc response at zero bias. This interpretation is supported by the THz near-field map and the absence or near-noise-floor THz response in the SSD control device. For a detailed discussion, please refer to the Supplementary Note 1. Finally, we evaluate the practical relevance of this response through black-body radiation measurements at room temperature (Supplementary Fig. S19). The SASD produces stable photocurrent signals under black-body illumination across a wide temperature range, with a nearly constant responsivity of approximately 0.5 A/W. These results further highlight the capability of cross-scale asymmetry to support ultrabroadband, zero-bias photodetection from the visible to the THz regime.

We also considered possible competing mechanisms, including PTE, bolometric response, plasmonic heating, contact-related photocurrent, and hot-electron injection. These effects may exist as background contributions in a microstructured device, but several observations argue against their dominance: the spatial decomposition shows a dominant unidirectional axial component rather than a reverse-signed contact/thermal component; temperature-dependent power measurements show no qualitative change from 80 to 210 K; the response exhibits microsecond-scale switching rather than slow device-scale thermal diffusion; and orientation-dependent devices show that the response is maximized only when the Te current axis matches the microstructure-induced optical-field asymmetry. Together, these results support the dominant role of symmetry-compatible optical-field redistribution rather than conventional thermal or interface-driven mechanisms.

Quantitative BPV coefficients and benchmarking across visible-IR-THz regimes

To place the performance of the SASD device in a broader context, we quantitatively benchmark its bulk photovoltaic response against representative BPVE-active materials reported across the visible, infrared, and terahertz spectral regimes. We focus on two commonly used figures of merit under zero bias: the short-circuit current density (J_sc) and the effective BPV coefficient extracted from the power-dependent response. We emphasize that the coefficient compared here is a device-level effective BPV coefficient, rather than an intrinsic shift-current tensor or intrinsic material parameter. The relatively strong SSD response reflects the high-quality non-centrosymmetric Te nanowires, whereas the SASD provides an additional device-level enhancement by improving the spatial overlap among local optical excitation, the Te crystal-allowed current direction, and the two-terminal collection direction.

Figure 5a compares the dependence of J_sc on incident optical power density for a range of materials operating at different wavelengths. For all systems, J_sc exhibits a linear scaling with power density, consistent with second-order nonlinear photoresponses. Under excitation at 638 and 1550 nm, the SASD device achieves short-circuit current densities that are among the highest reported for BPVE-based photodetectors operating without external bias, highlighting the effectiveness of synergistic asymmetry in enhancing zero-bias response. We also summarize reported short-circuit current densities of representative BPVE-active materials as a function of their operating wavelength (Supplementary Fig. S20).

Fig. 5. Quantitative benchmarking of device-level effective BPV response across measured visible-IR-THz bands.

a Short-circuit current density (J_sc) as a function of incident optical power density for representative BPVE-active materials under zero bias at different excitation wavelengths, showing linear scaling characteristic of second-order photoresponses. Data for materials: MoS₂(SG)⁶, WS₂(NTs)⁵, TTF-CA²⁶, PZTO²⁷, OMPH²⁸, MPI²⁹, KBNNO³⁰, BiFeO₃(BFO)³¹, BFCO³², and BaTiO₃(BTO)³³. b Comparison of device-level effective BPV coefficients as a function of wavelength. The plotted coefficients are effective device-level quantities extracted from measured zero-bias photocurrent density and include the combined effects of intrinsic material response, absorption, geometry, field redistribution, carrier transport, and collection efficiency. The SASD device exhibits measurable zero-bias responses across selected visible, infrared, and THz bands. Data for materials: 3R-MoS₂⁹, WS₂(NTs)⁵, MoS₂(SG)⁶, BP/WSe₂³, BiFeO₃(BFO)³¹, TaAs¹³, CuInP₂S₆(CIPS)¹¹, BaTiO₃(BTO)³³, strained Fe-doped LiNbO₃ (LNO: Fe(s))³⁴, are shown as open and solid symbols. The BFO value corresponds to β₂₂, the BTO value corresponds to β₃₁, and the other values are effective coefficients. NTs: nanotubes; BP: black phosphorus.

To further quantify the response strength across spectral regimes, we extract device-level effective BPV coefficients and compare them with literature values. This device-level role is distinct from conventional plasmonic-assisted photodetection, where metallic structures mainly enhance absorption, generate hot carriers, create thermal gradients, or assist interface rectification; here, the asymmetric microstructure is used primarily to redistribute the optical field in a symmetry-compatible manner with the Te crystal-allowed current direction (Supplementary Table 1). As shown in Fig. 5b, the SASD exhibits a measurable and sizable response spanning from the visible to the terahertz regime. The effective coefficients, estimated along the principal polarization axis using $J_{sc}={\beta }_{eff}I$, reach values of approximately 11.7 V⁻¹ at 638 nm, 0.70 V⁻¹ at 1550 nm, and 0.44 V⁻¹ at 0.033 THz. While direct comparison of BPV coefficients across different material systems requires caution due to variations in tensor definitions and experimental configurations, the data clearly indicate that the SASD maintains a comparatively strong response even as the photon energy is reduced into the THz regime, where BPVE signals are rarely observed.

Notably, this behavior contrasts with many previously reported BPVE systems, which typically exhibit pronounced responses only within limited spectral windows. The ability of the SASD to sustain a measurable and efficient zero-bias photoresponse from the visible to the THz regime reflects the role of cross-scale asymmetry in extending bulk photovoltaic functionality beyond material-intrinsic constraints. Long-term stability measurements further confirm that the device performance remains robust and reproducible over extended operation (Supplementary Fig. S21).

These benchmarking results demonstrate that coordinating intrinsic crystal asymmetry with engineered device-level asymmetry provides an effective route for realizing ultrabroadband bulk photovoltaic photodetection. Rather than maximizing a single material parameter, this approach enables a balanced enhancement of zero-bias photoresponse across multiple spectral regimes, offering a practical pathway toward energy-efficient photodetectors operating from the visible to the terahertz range.

This work demonstrates that a robust zero-bias bulk photovoltaic response can be sustained across a substantially broad spectral range when intrinsic crystal asymmetry is coherently combined with device-scale optical-field asymmetry. The central mechanism is not the amplification of a single microscopic process, but rather the effective rectification enabled by symmetry matching across length scales. In the SASD, the helical tellurium nanowire provides an intrinsic non-centrosymmetric response channel, while the asymmetric microstructure reshapes the electromagnetic excitation into a spatially polarized and inhomogeneous near-field. When the structure-defined driving direction aligns with a crystal-allowed current axis, optical excitation is more efficiently projected onto the permitted nonlinear response channel, leading to a substantially enhanced zero-bias dc output compared with symmetric control devices.

This cross-scale symmetry picture remains applicable across disparate photon-energy regimes. In the visible and infrared ranges, the enhanced photoresponse is consistent with bulk-photovoltaic and photogalvanic contributions strengthened by optical-field confinement and spatially redistributed absorption. As the photon energy is reduced far below the bandgap, interband transitions become strongly suppressed, and symmetric tellurium devices exhibit negligible response. In contrast, the SASD maintains a measurable room-temperature terahertz photoresponse under zero bias. In this low-energy regime, intraband carrier dynamics dominate optical absorption; however, the strong local field gradients and broken spatial symmetry imposed by the microstructure enable symmetry-allowed even-order rectification, giving rise to a net dc photocurrent. The agreement between near-field simulations, experimental field mapping, bias-independent current–voltage characteristics, and symmetric-device controls supports a unified interpretation based on symmetry-enabled rectification rather than junction-driven or purely photoconductive effects. Although demonstrated here in chiral tellurium, the concept is not restricted to a specific material system. Beyond the specific material system studied here, these results provide a complementary device-level route for extending bulk photovoltaic functionality beyond material-intrinsic limitations.

Methods

Synthesis of helical Te nanowires

Tellurium nanowires were synthesized via single-temperature-zone CVD¹². TeO₂ bulk source and Si/SiO₂ substrate were placed in a tube furnace; after H₂/Ar purging (30 sccm), the furnace was heated to 973–1023 K over 30 min and held for 15–25 min, followed by natural cooling. Nanowire morphology and crystallinity were characterized by optical microscopy (Nikon) and 532 nm micro-Raman spectroscopy, with helical Te NWs further inspected via Nikon LV100ND and WITec Alpha 300 (532 nm).

Fabrication of SASD and SSD devices

Two hundred eighty-five nanometers SiO₂/Si wafers were cleaned, dehydrated at 180 °C, spin-coated with AR-P 6200 photoresist, and baked. Ninety-nanometers-thick Au disk arrays with 10 nm Cr adhesion layer were patterned via EBL and lift-off; symmetric Au strip devices were fabricated for comparison. Te NWs were dry-transferred via PMMA to bridge electrodes, followed by acetone cleaning and annealing at 120 °C in Ar/H₂ (95:5) for 2 h to enhance contact.

Structural characterization

Surface morphology and height profiles were obtained with AFM (Bruker Dimension Icon). Spherical Aberration-Corrected Transmission Electron Microscope (AC-TEM) confirmed the chiral 1D lattice. Polarization-resolved second-harmonic generation (SHG) was measured using a femtosecond OPA (1064 nm pump, 520 nm SH) and a rotation stage to map azimuthal dependence. A commercial THz s-SNOM system (NanoTHz‑c, Miji Tech) was employed to characterize the terahertz near‑field localization.

Numerical simulations

Three-dimensional finite-difference time-domain (FDTD, Lumerical) simulations employed uniform mesh, periodic boundaries in x-y, and perfectly matched layers along z. The simulated structure consists of a silicon substrate, SiO₂ (285 nm thickness), Au electrode (10 nm Cr/ 90 nm Au), and air. The power density absorption was calculated using the equation: $P_{abs}=1/2\omega Im\left(\epsilon \right){\mid E\mid }^2$, where $\omega$ is the light frequency and $Im\left(\epsilon \right)$ is the imaginary part of the dielectric function. The field-ratio reference was calculated using two FDTD simulations with identical wavelength, polarization, boundary conditions, mesh settings, and sampling region. The SSD reference retained the strip-electrode/substrate geometry but removed the Au disk microstructure.

Electrical and photoresponse measurements

All measurements were performed in ambient conditions on a probe station. I–V characteristics were collected with a Keysight B2902A SMU (MStarter 200). Incident power was calibrated with a thermopile sensor (Ophir). Polarization dependence was recorded by inserting a half-wave plate. THz excitation was provided by a solid-state source (Toptica TeraScan; 0.02-0.22 THz), collimated; modulation was achieved via internal bias-tee (up to 10 kHz). Photocurrent transients were captured by a 200 MHz digital oscilloscope (PicoScope 5444D).

Scanning photocurrent mapping

For zero-bias scanning photocurrent mapping, a 1550 nm fibre-coupled laser was focused through a 50× objective (NA = 0.6) and raster-scanned by a galvo mirror system (ThorLabs GVS012). The pixel-by-pixel short-circuit current was synchronously sampled by a lock-in amplifier (Stanford SR830) referenced to a mechanical chopper (1.1 kHz).

Author contributions

G.H.L. and X.S.C. conceived the project. H.B.S. provided the photosensitive material. C.L.L., S.N. and X.L. designed the experiments. S.N., X.L., R.S.C., C.Y.P., F.Y.Z., X.H.F., H.B.G. and S.A.M. conducted all the experiments. G.H.L., C.L.L. and S.N. participated in writing the manuscript. G.H.L., C.L.L., S.N., X.L., L.W., M.H. and W.W.T. analyzed the data. All authors contributed to the organization of the manuscript.

Peer review

Peer review information

Nature Communications thanks Hua Wang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by National Key Research and Development Program of China (2023YFA1406900); National Natural Science Foundation of China (U2341226, 62550006, U25A203600, U23A6002, 12227901); Zhejiang Provincial Natural Science Foundation (LZ22F040003, LY22A040002, LZ24F050006); China National Postdoctoral Program for Innovative Talent (Grant No. BX20250352), and China Postdoctoral Science Foundation (Grant No. 2025M780543); Hangzhou West Science and Technology Innovation Corridor Youth Project and Youth Special Award, Open Fund of State Key Laboratory of Infrared Physics (SITP-NLIST-YB-2024-09, SITP-NLIST-YB-2023-13), Excellent Postdoctoral Research Projects of Zhejiang Province (ZJ2021019), the Research Funds of Hangzhou Institute for Advanced Study, UCAS (B02006C019025, B02006C021010).

Data availability

Relevant data supporting the key findings of this study are available in the article and Supplementary Information file. All raw data generated in this study are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77486-0.