时空分辨太赫兹发射纳米显微成像自旋电子光电流

Introduction

Gaining a detailed understanding of ultrafast charge and spin currents is vital for the development of next-generation technologies, from highly efficient solar cells and optimized light-driven chemical reactions to terahertz (THz) data-processing devices^1,2,3.

In this context, the femtosecond (fs) time and nanometer (nm) length scales are of particular importance. These spatiotemporal scales not only match the dimensions of modern electronic and spin-orbitronic devices that are envisioned to operate at ultrafast THz rates but also coincide with the native ranges that govern the relaxation of electronic systems in terms of energy, as well as linear and angular momentum⁴. Despite this, experimental approaches that combine such resolutions are scarce⁵, often providing only indirect insights through experimentally demanding setups^6,7,8. Consequently, there is an urgent need for probes capable of simultaneously capturing fs charge and spin transport at the nanoscale⁹.

An established tool to measure ultrafast spin and charge photocurrents with fs time resolution is far-field THz-emission spectroscopy (TES)¹⁰. In TES, a fs optical excitation pulse triggers a charge-current burst, which subsequently radiates a THz electric field into the far field. By analyzing the polarization and amplitude of this radiation, TES can characterize the current in magnitude and direction, thus providing a vectorial probe of the ultrafast carrier dynamics. However, the spatial resolution of TES is fundamentally limited by diffraction to the micrometer range^11,12, preventing the visualization of local charge-current distributions within or close to the excitation spot.

Recent advances have bypassed this diffraction limit by combining TES with near-field scanning probes, where a sharp metallic tip acts as a THz nano-antenna. In this THz-emission nanoscopy (TEN), the spatial resolution is limited only by the tip apex size—ideally reaching tens of nanometers¹³. This approach yielded remarkable local insights into ultrafast out-of-plane photocurrents in 2D materials¹⁴, semiconductors^15,16, and metallic nanostructures^17,18,19,20. However, a significant challenge remains: many fundamental ultrafast phenomena involve in-plane coupled electronic charge and spin dynamics. Because near-field probes are primarily sensitive to out-of-plane currents, it has remained unclear if and how TEN can provide spatiotemporal “filming” of these critical in-plane spin-driven charge dynamics on the nanoscale.

Here, we use TEN to address this challenge by utilizing a fiber-coupled spintronic THz emitter (STE)^21,22,23 as a model system, where an ultrafast light-induced spin current is converted into a spatially confined in-plane charge-current distribution. By laterally translating the fiber-coupled STE beneath a stationary metallic tip of a scattering near-field optical microscope, we map the propagation dynamics in real space. Surprisingly, the detected THz electric field spatially evolves in a dipolar-like pattern. Supported by finite-element simulations, we demonstrate that this TEN signal originates from the out-of-plane electric-field components close to the STE surface that inherently accompany in-plane charge currents in metallic films. These insights resolve the persisting ambiguity in the TEN community regarding the observation of ultrafast spin-related in-plane charge transport. In combination with the more established TEN-probing of THz out-of-plane currents, our findings pave the way for TEN to become a fully vectorial probe for the spatiotemporal imaging of ultrafast spin and charge dynamics on the nanoscale applicable to a wider class of material systems.

Results

Fiber-coupled spintronic THz-emission nanoscopy

To capture spin and charge currents with simultaneous fs and nm resolution, we integrate a fiber-tip spintronic terahertz emitter into a scattering near-field optical microscope (SNOM, attocube) (Fig. 1). This combination provides two important advantages: (i) the single-mode pump waveguiding guarantees a scan-position independent Gaussian excitation profile with sub-THz-wavelength dimensions, and (ii) the flexible-fiber coupling maintains a constant pump-probe delay and an invariant THz-detection response function throughout the scanning-probe motion.

Fiber-coupled spintronic terahertz emitter

Our localized THz source consists of an efficient tri-layer STE^21,22,23,24 that is sputter-deposited onto the facet of a single-mode glass fiber (Fig. 1a, b). The STE is composed of a metallic thin-film heterostructure with a central ferromagnetic layer (CoFeB, 1.8 nm thickness) and two adjacent heavy-metal layers with large but opposite spin-Hall angles (W and Pt, 2 nm thickness each). For THz generation from the STE, a femtosecond optical pulse incident from the fiber (Fig. 1c) excites the CoFeB layer, triggering ultrafast demagnetization and a concurrent spin-current injection into the heavy metal layers^24,25. Due to their large and opposite spin-Hall angles, these spin currents are converted into a transverse in-plane charge-current burst via the inverse spin-Hall effect inside the heavy metals, radiating a THz electromagnetic pulse (Fig. 1a).

Nanoscale spatiotemporal mapping

To perform non-local nanoscale imaging, the fiber-coupled STE is mounted to the SNOM’s 3-axis piezo stage and positioned underneath the sharp SNOM tip (Fig. 1c). The metallic SNOM tip is primarily sensitive to out-of-plane electric fields (\({E}_{z}\)), with its nanometer-scale apex locally enhancing near-fields by over an order of magnitude (see Supplementary information). By oscillating the SNOM tip above the sample at a frequency \(\Omega\), we extract the near-field contribution via demodulating the signal at harmonics \(n\Omega\) (refs. ^13,26). The emitted THz electric field is collimated by an off-axis parabolic mirror and focused onto a photoconductive antenna, allowing for time-domain sampling of the THz waveform (Fig. 1d, see the “Materials and methods” section).

Importantly, because the fiber-coupled path eliminates changes in the optical path length as it bends slightly throughout raster-scan movements, we can directly visualize the spatiotemporal evolution of the THz dynamics with a constant THz-detection response function at every scan position of the SNOM.

Experimental results

THz-emission nanoscopy movies from the fiber-coupled STE

We obtain THz-emission movies with nanometer spatial resolution by tracking the THz-emission signal as a function of the fiber-coupled-STE position beneath the SNOM tip (see the “Materials and methods” section). We record 3D datasets with 1 temporal and 2 spatial dimensions in a pump-probe fashion. Due to the fully fiber-coupled pump path, the temporal delay axes are identical for each pixel, eliminating any relative shift. Figure 2a shows six movie frames of the THz-emission signal, lock-in detected at the tip-oscillation frequency, for distinct detection-time delays. Intriguingly, the spatial pattern follows a clear dipolar pattern. This pattern is centered at the fiber core, which reveals a surprising result: no THz-emission signal is detected at the center of the Gaussian STE-excitation spot where the pump intensity and, thus, the current density is maximum. Instead, the off-centered dipole minima and maxima oscillate over time, with the dipole axis oriented perpendicularly to the in-plane STE magnetization (\({{\bf{M}}}\)). This dipole-to-magnetization orientation has also been confirmed in a previous spintronic TEN study²⁰.

To understand these dipole features, we first recall that the SNOM metallic tip selectively enhances the out-of-plane electric-field component (\({E}_{{{\rm{z}}}}\)). We combine this with the analytical solution for the electric-field distribution in the near field surrounding a point-like in-plane electric dipole embedded in a thin metallic film extending to infinity in the plane²⁷. The pattern of the out-of-plane electric field \({E}_{{{\rm{z}}}}\) for a delta-like in-plane dipole density follows:

$${E}_{{{\rm{z}}}}\propto \frac{1}{{r}^{2}}\cos \varPhi$$

(1)

Here, \(r\) is the distance from the dipole center and \(\varPhi\) is the angle in the sample plane with respect to the dipole axis. Equation (1) describes a two-lobe structure with a center zero line that decays according to \(1/{r}^{2}\), qualitatively matching our experimental results (Fig. 2). Consequently, we conclude that the spintronic TEN images are the \({E}_{{{\rm{z}}}}\) signature of the in-plane charge-current density driven by the laser-excited STE.

Based on an inverse-spin-Hall-effect driven spin-to-charge current conversion²², the transient charge-current dipole is expected to be oriented orthogonally to the STE’s in-plane magnetization \({{\bf{M}}}\) (Fig. 1a). In our measurement (Fig. 2), with \({{\bf{M}}}\) parallel to the x-axis, the observed dipole axis is indeed orthogonal to the direction of \({{\bf{M}}}\). This agreement bolsters our interpretation of the spintronic TEN image contrast and aligns with related SNOM studies of in-plane electric dipoles^28,29,30.

Furthermore, this mechanism also explains why previous TEN studies^19,20 could obtain THz-emission signals from STEs despite the supposed lack of in-plane sensitivity: in those works, where femtosecond excitation is realized via free-space sample excitation, the THz-emission signal was likely optimized by unintentionally moving the excitation-laser spot relative to the SNOM tip. This created an off-centered pump-tip configuration, allowing the tip to pick up the finite out-of-plane electric field components at the edges of the in-plane charge-current distribution.

Temporal evolution and propagation

We continue our analysis by extracting the THz signal for a given scanning position. A clear oscillating waveform with subpicosecond dynamics vs. detection-time delay emerges, typical of STEs. Remarkably, a propagation delay becomes visible when comparing THz waveforms at locations 4 and 14 μm away from the emitter center (Fig. 2a). We tentatively assign this to the velocity of the surface plasmon polariton moving along the air/metal interface³¹. This propagating nature is even more apparent in the full spintronic TEN movie (see https://doi.org/10.5281/zenodo.18739933) and will be analyzed in more detail later.

Higher-harmonic analysis

As shown in Fig. 2b, the spintronic TEN images recorded with higher-harmonic demodulation (1st–4th) follow identical spatial patterns with decreasing signal-to-noise ratios. While higher-harmonic demodulation typically increases the spatial resolution by restricting the interaction window to the moment the SNOM tip is closest to the STE¹³, our results suggest we are already resolving the dominant spatial features at the first harmonic with our 1-µm experimental scan step size. The spintronic TEN images are dominated by the long-range \(1/{r}^{2}\) decay (see Eq. (1)). Notably, nanometer features, such as the polarity flip at the dipole center, can still be resolved.

Frequency-domain analysis

After concluding our time-domain analysis, we now turn our attention to the frequency domain. Figure 3a shows the spectral amplitude and spectral phase obtained by Fourier transforming the time-domain signal at the spatial location with the largest peak-to-peak amplitude (see gray arrow). The obtained center frequency of approximately \(\omega /2\pi =1\,{{\rm{THz}}}\) is consistent with the laser pulse duration of 70 fs and the THz-detector sensitivity. The spectral phase is relatively flat, indicating a minor chirp of the THz waveform as typical for STEs²².

The spectral-amplitude map at a frequency of 1 THz (Fig. 3b) reinforces the dipolar spatial profile observed in the time domain (Fig. 2). A slight asymmetry along the y-axis is evident, likely originating from a minor change in tip-sample distance during the scan. In the spectral-phase map (Fig. 3c), we observe an abrupt, nanoscale \(\pi\) phase jump at \(y=0\), as predicted by the sign change of the \(\cos \varPhi\) in Eq. (1). Additionally, the spectral phase exhibits a constant radial gradient that switches sign at \(y=0\). This phase gradient is the frequency-domain signature of a propagation delay observed in the time domain (Fig. 2a).

Note that to increase the signal-to-noise ratio, all images obtained for different harmonic demodulations shown in Fig. 2b are summed up prior to Fourier transformation. Minor diagonal stripes in the phase map in Fig. 3c arise from periodic instabilities of the SNOM during the extended measurement duration (see the “Materials and methods” section).

Finite-element simulations

Simulating the spintronic TEN experiment

While Eq. (1) provides a qualitative understanding of the observed spintronic TEN image contrast under idealized conditions, a rigorous description requires detailed modeling of the experimental geometry. We account for the finite size of the fiber-coupled STE and the complex photonic environment using a finite element solver (COMSOL Multiphysics, RF module). We simulate the experimental geometry, including the fiber facet, thin metallic layer with a 6 nm thickness and scanning probe, driven by a Gaussian-shaped in-plane electric charge-current density, reflecting the fiber-mode symmetry. From this, we extract the out-of-plane electric field component \({E}_{z}\) at 1 THz (see the “Materials and methods” section).

Comparison between simulation and experiment

In Fig. 4, we present a back-to-back comparison of the simulation results and experimental measurements. The THz-spectral-amplitude maps at 1 THz for scan step sizes of 1 µm and 300 nm show remarkably good agreement with the simulation (Fig. 4a). This consistency further substantiates our interpretation: the TEN signal arises from the local E_z components generated by a spatially confined in-plane current source.

The agreement is further exemplified by comparing line scans at \(x=0\,{{\rm{\mu }}}{{\rm{m}}}\) (Fig. 4b), which reveal an excellent match between the measured and simulated peak positions, as well as the center-dip shape with a 1/e² width of ~400 nm. We note that the measured line-cut exhibits a slightly shallower spatial falloff than the simulation. This broadening likely indicates a deviation from a purely Gaussian current profile. In our simulation, we assume a Gaussian profile in line with the fiber mode profile. However, at the given experimental pump conditions, fiber-tip STEs potentially exhibit saturation at the center of the excitation spot³². This could result in a “flat-top” current profile, possibly leading to the observed spatial broadening.

We also extract the spectral phase along the line cut at \(x=0\,{{\rm{\mu }}}{{\rm{m}}}\) (Fig. 4c), which follows a clear linear trend in both experiment and simulation. Linear fitting of the simulation data yields a phase gradient along the y direction of \(\pm 0.011{{\rm{\pi }}}{{\rm{\mu }}}{{{\rm{m}}}}^{-1}\), corresponding to a subluminal 1-THz phase velocity of about 60% of the speed of light. Such subluminal velocities have been reported for similar surface plasmon-polariton modes³³.

Optimizing experimental conditions

To explore the robustness of the presented technique, we perform a series of simulations varying the spatial extent of the in-plane-current distribution (Fig. 5a) and the average metal-film conductivity (Fig. 5b). In these simulations, the initial peak charge-current density j₀ is fixed at 1 A/m² (see the “Materials and methods” section).

Importantly, our single-mode fiber-tip STE design, with a pump mode-field diameter (MFD) of 10.5 µm, is near-optimal for efficient \({E}_{z}\) generation. For MFD values exceeding 25 μm, the finite geometry limited by the fiber diameter of 125 μm begins to interfere with the spatial pattern of \({E}_{z}\), as the fields can no longer be considered decaying in an infinite plane. This confirms that Eq. (1) remains valid only when the current source is much smaller than the underlying geometry and far away from physical boundaries.

Interestingly, we find that the average STE metal-film conductivity primarily scales the amplitude of \({E}_{z}\) without altering its spatial distribution (Fig. 5b). For conductivities exceeding 10⁴S/m, the \({E}_{z}\) amplitude starts saturating, making the TEN signal robust against moderate conductivity fluctuations in the experimentally relevant region around 10⁶ S/m^22,34. This stability simplifies the interpretation of spintronic TEN images significantly, as spatial variations in film quality do not require complex correction factors. Conversely, it also highlights that the TEN signal strength will decrease significantly when investigating nonmetallic or poorly conducting samples.

Discussion

To contextualize our results, we note that our fiber-coupled TEN system offers significant advantages over state-of-the-art TEN approaches. In standard, non-fiber-coupled SNOM setups, one typically encounters a trade-off: when the tip is moved while keeping the sample and excitation spot stationary, the setup-response function changes severely as the tip position varies with respect to the THz-detection unit³⁵. Alternatively, when laterally translating the sample under a stationary tip, the setup-response function remains constant, but instead changes in optical pump-path length, excitation-profile and -position on the sample complicate image interpretation. Conversely, our fiber-coupled setup ensures invariance for all three during scanning: invariant excitation profile and -position, constant THz-detection delay, and a constant setup-response function. This enables real-space visualization of out-of-plane and in-plane currents with simultaneous femtosecond and nanometer resolution.

In conclusion, we film the in-plane spin-driven charge dynamics in a fiber-coupled spintronic terahertz emitter in real space. Our simulations reproduce the measured near-field spatiotemporal maps, providing access to the fundamental image-forming mechanism behind spintronic terahertz-emission nanoscopy: the measured signal originates from the out-of-plane electric-field component that inevitably accompanies localized in-plane current distributions. This resolves an ambiguity in the TEN community and explains why previous studies could detect in-plane currents despite the probe’s inherent vertical sensitivity.

We foresee broad applications of this table-top technique in heterogeneous material systems and nanostructures, including the study of magnetic nanodomains in altermagnets³⁶ or 2D magnets³⁷. Beyond magnetism, TEN is now equipped to study photocurrent heterogeneities in a wide range of systems on their native spatiotemporal scales.

Materials and methods

Details of measurements

For the measurements visualized in Figs. 2, 3, an area of 30 × 30 µm² is raster-scanned by moving the fiber-coupled STE with a step size of 1 µm. Per pixel, three THz waveforms are recorded, each sampled in 85-fs steps over a 3-ps delay window, with a lock-in time constant of 800 ms. These three waveforms are then averaged to yield a single waveform per pixel acquired over a total acquisition time of 84 s. To achieve higher spatial resolution (see zoom in Fig. 4), the raster-scan area is reduced to 10 × 10 µm², and the measurement is repeated with a step size of 300 nm.

The fiber-tip STE is pumped by an Erbium-doped fiber-laser system with a repetition rate of 100 MHz (Menlo ELMO and Menlo ELMA). The average pump power is set to 2 mW using a variable optical attenuator positioned just before the fiber-tip STE. The total pump-fiber length is chosen such that the pump-pulse duration reaches its Fourier limit at the fiber-tip STE (70 fs FWHM). The 1/e² mode-field diameter of the used single-mode pump fiber is specified as 10.5 µm. A detailed illustration of the experimental setup is shown in Fig. S1.

Correction of periodic instrument drifts

Over the measurement duration of ~21 h, a periodic drift of the instrument was evident across several independent signal channels. It was most pronounced in the z-axis channel of the piezo stage and likely caused by temperature fluctuations within the laboratory. These instabilities caused the acquired waveforms to periodically drift in temporal delay (see Fig. S5). To correct this drift, a sinusoidal two-dimensional delay-correction map parametrized by a single spatial frequency, phase, and amplitude is defined. The frequency and phase are obtained by applying a spatial Fourier transform to the two-dimensional map of the scalar z channel (z-axis of the piezo stage) and selecting the spectral-peak value. The amplitude is set to 56 fs, which minimizes the spatial distortion of the THz signal at a fixed delay. The delay-correction map then defines a delay shift for each pixel that is subsequently applied to the measurement data.

Details of the finite-element simulation

We use COMSOL Multiphysics to simulate THz emission from a fiber-tip spintronic THz emitter (STE). The primary charge current is assumed to be proportional to the absorbed pump energy density. To isolate purely electromagnetic effects and exclude the spintronic contribution in the simulations, we prescribe a primary charge current inside a homogeneous metallic film³⁴ (thickness \(d=6\,{{\rm{nm}}}\) and conductivity \(\sigma =1\times {10}^{6}\,{{\rm{S}}}/{{\rm{m}}}\)) as \({j}_{{{\rm{primary}}}}\left(r\right)={j}_{0}\exp (-2{r}^{2}/{w}^{2})\) where \(r\) is the radius, \(w\) is the beam waist (with \({{\rm{MFD}}}=2w\)), and \({j}_{0}\) is the amplitude of the primary current density. In all simulations presented in this work, we fix \({j}_{0}=1\,{{\rm{A}}}/{{{\rm{m}}}}^{2}\), independent of mode-field diameter, film thickness, and electrical conductivity.

We map both the in-plane and out-of-plane electric fields and charge-current distributions, as shown in the “Results” section and the Supplementary information. We observe that the in-plane electric field is continuous across the film–air interface, whereas the out-of-plane component vanishes inside the metallic film and exhibits a discontinuity at the interface, becoming nonzero in the adjacent air region. Accordingly, all out-of-plane electric-field results presented in this paper correspond to the field evaluated immediately outside the metallic interface in air.

Data availability

The data that support the findings of this study are available at https://doi.org/10.5281/zenodo.18739933. Additional raw data are available from the corresponding authors upon reasonable request.

References

1. Walowski, J. & Münzenberg, M. Perspective: ultrafast magnetism and THz spintronics. J. Appl. Phys. 120, 140901. https://doi.org/10.1063/1.4958846 (2016).

   Article  ADS  Google Scholar 

2. Ponseca, C. S. et al. Ultrafast electron dynamics in solar energy conversion. Chem. Rev. 117, 10940–11024 (2017).

   Article  Google Scholar 

3. Carlin, C. C. et al. Nanoscale and ultrafast in situ techniques to probe plasmon photocatalysis. Chem. Phys. Rev. 4, 041309 (2023).

   Article  Google Scholar 

4. Kampfrath, T. et al. Ultrafast and terahertz spintronics: Guest editorial. Appl. Phys. Lett. 123, 050401. https://doi.org/10.1063/5.0167151 (2023).

   Article  ADS  Google Scholar 

5. Yao, Z. H. et al. Nanoimaging and nanospectroscopy of polaritons with time resolved s-SNOM. Adv. Opt. Mater. 8, 1901042 (2020).

   Article  Google Scholar 

6. Müller, M., Paarmann, A. & Ernstorfer, R. Femtosecond electrons probing currents and atomic structure in nanomaterials. Nat. Commun. 5, 5292 (2014).

   Article  ADS  Google Scholar 

7. Zayko, S. et al. Ultrafast high-harmonic nanoscopy of magnetization dynamics. Nat. Commun. 12, 6337 (2021).

   Article  ADS  Google Scholar 

8. Von Korff Schmising, C. et al. Imaging ultrafast demagnetization dynamics after a spatially localized optical excitation. Phys. Rev. Lett. 112, 217203 (2014).

   Article  ADS  Google Scholar 

9. Guo, X. et al. Terahertz nanoscopy: advances, challenges, and the road ahead. Appl. Phys. Rev. 11, 021306 (2024).

   Article  ADS  Google Scholar 

10. Koch, M. et al. Terahertz time-domain spectroscopy. Nat. Rev. Methods Prim. 3, 48 (2023).

    Article  Google Scholar 

11. Murakami, H. & Tonouchi, M. Laser terahertz emission microscopy. C. R. Phys. 9, 169–183, https://doi.org/10.1016/j.crhy.2007.07.010 (2008).

    Article  ADS  Google Scholar 

12. Von Hoegen, A. et al. Imaging a terahertz superfluid plasmon in a two-dimensional superconductor. Nature 650, 869–874, https://doi.org/10.1038/s41586-025-10082-2 (2026).

    Article  ADS  Google Scholar 

13. Ocelic, N., Huber, A. & Hillenbrand, R. Pseudoheterodyne detection for background-free near-field spectroscopy. Appl. Phys. Lett. 89, 101124. https://doi.org/10.1063/1.2348781 (2006).

    Article  ADS  Google Scholar 

14. Plankl, M. et al. Subcycle contact-free nanoscopy of ultrafast interlayer transport in atomically thin heterostructures. Nat. Photonics 15, 594–600, https://doi.org/10.1038/s41566-021-00813-y (2021).

    Article  ADS  Google Scholar 

15. Yao, Z. H. et al. Photo-induced terahertz near-field dynamics of graphene/InAs heterostructures. Opt. Express 27, 13611–13623, https://doi.org/10.1364/OE.27.013611 (2019).

    Article  ADS  Google Scholar 

16. Pizzuto, A., Ma, P. C. & Mittleman, D. M. Near-field terahertz nonlinear optics with blue light. Light Sci. Appl. 12, 96 (2023).

    Article  ADS  Google Scholar 

17. Klarskov, P. et al. Nanoscale laser terahertz emission microscopy. ACS Photonics 4, 2676–2680, https://doi.org/10.1021/acsphotonics.7b00870 (2017).

    Article  Google Scholar 

18. Pizzuto, A., Mittleman, D. M. & Klarskov, P. Laser THz emission nanoscopy and THz nanoscopy. Opt. Express 28, 18778–18789, https://doi.org/10.1364/OE.382130 (2020).

    Article  ADS  Google Scholar 

19. Li, P. Y. et al. Laser terahertz emission microscopy of nanostructured spintronic emitters. Appl. Phys. Lett. 120, 201102 (2022).

    Article  ADS  Google Scholar 

20. Cai, J. H. et al. Terahertz spin currents resolved with nanometer spatial resolution. Appl. Phys. Rev. 10, 041414 (2023).

    Article  ADS  Google Scholar 

21. Paries, F. et al. Fiber-tip spintronic terahertz emitters. Opt. Express 31, 30884–30893, https://doi.org/10.1364/OE.494623 (2023).

    Article  ADS  Google Scholar 

22. Seifert, T. et al. Efficient metallic spintronic emitters of ultrabroadband terahertz radiation. Nat. Photonics 10, 483–488, https://doi.org/10.1038/nphoton.2016.91 (2016).

    Article  ADS  Google Scholar 

23. Seifert, T. S. et al. Spintronic sources of ultrashort terahertz electromagnetic pulses. Appl. Phys. Lett. 120, 180401 (2022).

    Article  ADS  Google Scholar 

24. Rouzegar, R. et al. Laser-induced terahertz spin transport in magnetic nanostructures arises from the same force as ultrafast demagnetization. Phys. Rev. B 106, 144427 (2022).

    Article  ADS  Google Scholar 

25. Rouzegar, R. et al. Femtosecond signatures of optically induced magnons before ultrafast demagnetization. Preprint at https://arxiv.org/abs/2507.01796 (2025).

26. Hillenbrand, R. & Keilmann, F. Complex optical constants on a subwavelength scale. Phys. Rev. Lett. 85, 3029–3032, https://doi.org/10.1103/PhysRevLett.85.3029 (2000).

    Article  ADS  Google Scholar 

27. Margetis, D. & Luskin, M. On solutions of Maxwell’s equations with dipole sources over a thin conducting film. J. Math. Phys. 57, 042903 (2016).

    Article  ADS  MathSciNet  Google Scholar 

28. Niehues, I. et al. Nanoscale resolved mapping of the dipole emission of hBN color centers with a scattering-type scanning near-field optical microscope. Nanophotonics 14, 335–342 (2025).

    Article  ADS  Google Scholar 

29. Hillenbrand, R. et al. Coherent imaging of nanoscale plasmon patterns with a carbon nanotube optical probe. Appl. Phys. Lett. 83, 368–370 (2003).

    Article  ADS  Google Scholar 

30. Park, K. D. & Raschke, M. B. Polarization control with plasmonic antenna tips: a universal approach to optical nanocrystallography and vector-field imaging. Nano Lett. 18, 2912–2917, https://doi.org/10.1021/acs.nanolett.8b00108 (2018).

    Article  ADS  Google Scholar 

31. Zayats, A. V., Smolyaninov, I. I. & Maradudin, A. A. Nano-optics of surface plasmon polaritons. Phys. Rep. 408, 131–314 (2005).

    Article  ADS  Google Scholar 

32. Paries, F. et al. Optical damage thresholds of single-mode fiber-tip spintronic terahertz emitters. Opt. Express 32, 24826–24838 (2024).

    Article  ADS  Google Scholar 

33. Lummen, T. T. A. et al. Imaging and controlling plasmonic interference fields at buried interfaces. Nat. Commun. 7, 13156 (2016).

    Article  ADS  Google Scholar 

34. Rouzegar, R. et al. Broadband spintronic terahertz source with peak electric fields exceeding 1.5 MV/cm. Phys. Rev. Appl. 19, 034018. https://doi.org/10.1103/PhysRevApplied.19.034018 (2023).

    Article  ADS  Google Scholar 

35. Pizzuto, A. et al. Nonlocal time-resolved terahertz spectroscopy in the near field. ACS Photonics 8, 2904–2911 (2021).

    Article  Google Scholar 

36. Šmejkal, L., Sinova, J. & Jungwirth, T. Emerging research landscape of altermagnetism. Phys. Rev. X 12, 040501. https://doi.org/10.1103/PhysRevX.12.040501 (2022).

    Article  Google Scholar 

37. Grubišić-Čabo, A. et al. Roadmap on quantum magnetic materials. 2D Mater. 12, 031501 (2025).

    Article  Google Scholar 

Download references

Acknowledgements

We thank Tobias Kampfrath, Ioachim Pupeza, Maximilian Högner, Jean-François Lampin, and Pierre Koleják for valuable discussions and their general support. We gratefully acknowledge financial support from the German research foundation (DFG, Deutsche Forschungsgemeinschaft) through the collaborative research centers “SPIN + X” (268565370, TRR 173, project B11) and “Ultrafast spin dynamics” (328545488, TRR 227, project A05), and through the priority program “INTEREST” (468511389, SPP 2413, project ITISA2); from the European Research Council (101088303); from the Chinese Scientific Research Innovation Capability Support Project for Young Faculty (ZYGXQNJSKYCXNLZCXM-I3); and from the US National Science Foundation (ECCS-2300152). Moreover, we thank the research infrastructure SupraFAB (www.suprafab.fu-berlin.de), the France 2030 investment plan (PEPR SPIN ANR-22-EXSP 0003 - TOAST and ANR-22-EXSP-0007 - SPINTMAT), and the RENATECH Network for support.

Funding

Open Access funding enabled and organized by Projekt DEAL.

Author information

Author notes

1. These authors contributed equally: Felix Paries, Reza Rouzegar.

Authors and Affiliations

1. Department of Physics and Research Center OPTIMAS, RPTU Kaiserslautern-Landau, 67663, Kaiserslautern, Germany

   Felix Paries, Felix Selz & Georg von Freymann

2. Fraunhofer Institute for Industrial Mathematics ITWM, 67663, Kaiserslautern, Germany

   Felix Paries, Felix Selz, Daniel Molter & Georg von Freymann

3. Department of Physics, Freie Universität Berlin, 14195, Berlin, Germany

   Reza Rouzegar & Tom S. Seifert

4. Hangzhou International Innovation Institute, Beihang University, Hangzhou, 311115, China

   Jiahua Cai, Mingcong Dai & Xiaojun Wu

5. School of Electronic and Information Engineering, Beihang University, Beijing, 100191, China

   Jiahua Cai, Mingcong Dai & Xiaojun Wu

6. School of Engineering, Brown University, Providence, RI, 02912, USA

   Felix Selz, Johanna Koelbel & Daniel M. Mittleman

7. University of Lille, CNRS, Centrale Lille, Université Polytechnique Hauts-de-France, UMR 8520 - IEMN, 59000, Lille, France

   Geoffrey Lezier & Nicolas Tiercelin

8. Zhangjiang Laboratory, 100 Haike Road, Shanghai, 201204, China

   Xiaojun Wu

Authors

1. Felix Paries
2. Reza Rouzegar
3. Jiahua Cai
4. Mingcong Dai
5. Felix Selz
6. Johanna Koelbel
7. Geoffrey Lezier
8. Nicolas Tiercelin
9. Daniel Molter
10. Daniel M. Mittleman
11. Georg von Freymann
12. Xiaojun Wu
13. Tom S. Seifert

Contributions

F.P. conceived the idea and initiated the project, with facilitation by D.M., G.v.F., D.Mi., X.W., and T.S.S. F.P. led the experiments, conducted by F.P., J.C., M.D., F.S., and J.K. R.R. provided the theoretical framework and designed and performed all numerical simulations under the guidance of T.S.S. F.P. processed the measurement data. F.P., R.R., and T.S.S. interpreted the data. F.P., F.S., G.L., and N.T. fabricated the samples. F.P., R.R., and T.S.S. wrote the manuscript and created the figures. All authors contributed to revisions. The project was supervised by D.M., N.T., G.v.F., D.Mi., X.W., and T.S.S.

Corresponding authors

Correspondence to Felix Paries, Xiaojun Wu or Tom S. Seifert.

Ethics declarations

Conflict of interest

The authors declare no competing interests.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Paries, F., Rouzegar, R., Cai, J. et al. Spatiotemporal terahertz emission nanoscopy of spintronic photocurrent. Light Sci Appl 15, 386 (2026). https://doi.org/10.1038/s41377-026-02468-2

Download citation

• Received: 20 May 2026

• Revised: 04 August 2026

• Accepted: 17 August 2026

• Published: 08 October 2026

• Version of record: 08 October 2026

• DOI: https://doi.org/10.1038/s41377-026-02468-2