Terahertz spin-current transparency through rough interfaces
Jiří Jechumtál, Jakub Zázvorka, Ondřej Novák, Martin Rejhon, Peter Kubaščík, Lukáš Nowak, Petr Němec, Eva Schmoranzerová, Martin Veis, Lukáš Nádvorník, Zdeněk Kašpar · 2026
Read Terahertz spin-current transparency through rough interfaces by Jiří Jechumtál, Jakub Zázvorka, Ondřej Novák, Martin Rejhon, Peter Kubaščík, Lukáš Nowak, Petr Němec, Eva Schmoranzerová, Martin Veis, Lukáš Nádvorník, Zdeněk Kašpar (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-26837.txt.
Introduction
In their 2026 paper, Jechumtál and ten colleagues test how interface roughness affects spin-current flow in cobalt-platinum stacks, using terahertz emission spectroscopy. Roughness is tuned via a gold buffer layer while other growth settings stay fixed. Surprisingly, spin-current transparency falls by only about 30 percent even as roughness and grain size triple, with no measurable change in the terahertz spectrum. Spin transport, they conclude, is robust to imperfect interfaces on ultrafast timescales.
Catalog record
| CREATOR | Jiří Jechumtál, Jakub Zázvorka, Ondřej Novák, Martin Rejhon, Peter Kubaščík, Lukáš Nowak, Petr Němec, Eva Schmoranzerová, Martin Veis, Lukáš Nádvorník, Zdeněk Kašpar |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24897 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.26837
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Terahertz spin-current transparency through rough interfaces Jiří Jechumtál, Jakub Zázvorka, Ondřej Novák, Martin Rejhon, Peter Kubaščík, Lukáš Nowak, Petr Němec, Eva Schmoranzerová, Martin Veis, Lukáš Nádvorník, Zdeněk Kašpar Abstract Spin transport across interfaces is critical for spintronic devices, yet remains difficult to probe on ultrafast timescales. We use terahertz emission spectroscopy on Co | | Pt heterostructures whose interface roughness is tuned through the thickness of an underlying Au buffer layer, while leaving other growth parameters unchanged. From the measured THz electric field, we extract the interface spin-current transparency t s t_{s} after correcting for the changes in sample impedance and optical absorption of the stack. Surprisingly, we find that t s t_{s} decreases by only ∼ \sim 30% as the interface root-mean-square roughness and the lateral grain size both increase by a factor of three, with no measurable change in the THz spectrum. These results demonstrate that interfacial spin transport is relatively robust against morphological variations on ultrafast timescales, establishing terahertz emission spectroscopy as a reliable probe of spin dynamics across imperfect interfaces. I Introduction Interfacial properties are the key parameters in proof-of-concept 38 ; 36 ; 11 as well as industrially produced 21 spintronic devices, where they directly affect spin-pumping efficiency. A central role is played by the interface spin transparency t_{s} , defined as the fraction of spin current that successfully traverses the interface. This quantity is reduced by mechanisms such as interfacial spin memory loss 2 . Understanding and controlling t_{s} is therefore essential for device performance and for the development of new spintronic concepts. While t_{s} has been extensively studied in the DC and GHz regimes 2 ; 5 ; 9 ; 24 ; 7 ; 40 ; 15 and partially by all-optical methods 17 ; 22 , much less is known about how interface properties influence spin transport at THz frequencies. This becomes particularly relevant given the ongoing shift of spintronics toward ultrafast time scales, especially by the rapid advancements in antiferromagnetic 10 ; 3 and altermagnetic 31 spintronics. To experimentally probe the ultrafast spin currents and their penetration through the interface, the spintronic terahertz emitter (STE) 28 provides an ideal tool 8 . During the rapid development of STEs, and the simultaneous deeper understanding of the underlying physics 29 , many studies focused on improving STE performance by modifying the interface properties 20 ; 12 ; 37 ; 26 ; 34 . These studies, however, simultaneously modified multiple parameters (such as growth conditions, interlayer composition, defect density), making it difficult to isolate the role of interface morphology itself. As a result, an experiment isolating solely the role of interface roughness on spin transparency in the ultrafast regime is still missing. In this work, we investigate the terahertz (THz) pulse emission originating in sub-picosecond spin-current pumping through a ferromagnet | heavy metal (FM | HM) interface of variable roughness. By analyzing the THz pulse strength, sample conductance and optical absorptance of the FM layer, we can directly see the role of interface roughness on changing spin transparency t_{s} . We observe no dependence of the THz pulse spectrum on interface roughness and only a \sim 30% decrease of t_{s} despite the roughness root mean square (RMS) and the equivalent disk radius (EDR) varying by roughly a factor of 3. This finding is favorable for THz spintronic devices, since spin transparency remains relatively robust against a significant increase in interface roughness. Highly efficient operation can therefore be maintained without intensive interface engineering, while still leaving room for fine-tuning through morphological control. II Experimental Methods II.1 Spintronic THz emission and spin transparency extraction Our general approach to generate ultrafast spin-current flowing through an interface of variable roughness is shown in Fig. 1 . The sample stack consists of FM | HM | Au( d ), where d indicates the variable thickness of the Au buffer layer, which affects its roughness and is translated also in the FM | HM interface roughness 35 . First, the femtosecond laser pulse excites the sample stack and transfers a fraction of the incident optical power A_{\text{FM}}(d) into the FM layer, depending on the thickness d , due to propagation of light in layered media. This generates an ultrafast spin voltage in FM 29 ; 25 and consequently an ultrafast spin current pulse j_{s}(d) traversing through the interface with spin transparency t_{s} into HM layer. The amplitude of spin voltage and j_{s} is assumed to scale with A_{\text{FM}} and t_{s} 8 . Second, the out-of-plane spin-current pulse in HM is converted into the in-plane charge-current j_{c}\propto\gamma_{\text{SH}}j_{s} via the ISHE, where \gamma_{\text{SH}} is the spin-Hall angle in HM. Finally, the j_{c} emits a THz pulse with electric field E_{\text{THz}} right behind the sample 28 ; 27 : \begin{split}E_{\text{THz}}(d)=\frac{eZ_{0}}{n_{1}+n_{2}+Z_{0}G(d)}j_{c}(d)\propto\\ t_{s}(d)A_{\text{FM}}(d)Z(d)M_{\text{FM}}\gamma_{\text{SH}}\lambda_{\text{HM}}\tanh{\frac{d_{\text{HM}}}{2\lambda_{\text{HM}}}}\end{split} (1) where the first equality means that the emitted THz field relates to the in-plane charge current j_{c} via the generalized impedance of the sample, where e is the elemental charge, Z_{0}=1/(\epsilon_{0}c)\approx 377\,\Omega is the vacuum impedance, n_{1},n_{2} are refractive indices of sample silicon substrate ( \approx 3.4 ) and the surrounding atmosphere ( \approx 1 ), and G(d) is the total conductance of the whole sample stack. The second relation corresponds to the mechanism of j_{c} generation and includes magnetization of FM layer M_{\text{FM}} and the factor \lambda_{\text{HM}}\tanh\frac{d_{\text{HM}}}{2\lambda_{\text{HM}}} coming from the integration of j_{s} in heavy metal with thickness d_{\text{HM}} and spin current relaxation length \lambda_{\text{HM}} 27 . Relation \frac{eZ_{0}}{n_{1}+n_{2}+Z_{0}G(d)}=Z(d) for generalized impedance was used. From Eq. 1 , we see that the spin transparency t_{s}(d) is directly proportional to E_{\text{THz}}(d) and we can observe solely the role of buffer Au layer thickness d on t_{s}(d) just by renormalizing E_{\text{THz}}(d) on changing impedance Z(d) and optical energy absorbed in FM A_{\text{FM}}(d) . In the experiment, we measure the electro-optical THz signal EOS_{\text{THz}} , which is related to E_{\text{THz}} at the sample surface via the convolution with the setup function: EOS_{\text{THz}}(t)=[H*E_{\text{THz}}](t) . Owing to the fact that H(t) is not changing within our experiment, for spectrally identical signals, we can relate the strength of EOS_{\text{THz}} with t_{s}(d) in the following way: S_{\text{THz}}(d)\propto E_{\text{THz}}(d)\propto t_{s}(d)A_{\text{FM}}(d)Z(d) (2) II.2 Sample set and its growth The impact of Au layer thickness d and the induced roughness of the FM | HM interface is characterized by an atomic force microscope (AFM) on a calibration sample without the top FM layer. Two cases shown in Fig. 1 schematically represent samples without an Au layer and with a relatively thick and rough Au buffer layer. The influence of interfacial roughness on spin transparency was studied using two sets of samples. THz measurements were performed on Au(5) | Co(2) | Pt(5) | Au( d ) | Si, with the thickness of the gold buffer layer ranging from d= 0 to d= 20 nm. Platinum is chosen as the HM because of the very efficient spin-to-charge conversion by the strong (I)SHE with \gamma_{\text{SH}}\sim 10 % 30 ; 33 , which causes the conversion to be dominated solely in Platinum. The platinum crystalline quality and consequently its spin hall angle \gamma_{\text{SH}} is assumed to be the same across the whole sample set, due to constant deposition conditions. Part of the sample area was shielded in the sputtering process and reveals the bare substrate,