---
title: Optical Intersite Spin Transfer
url: https://www.emergentmind.com/topics/optical-intersite-spin-transfer-oistr
type: topic
---

# Optical Intersite Spin Transfer

Searching arXiv for recent papers on Optical Intersite Spin Transfer (OISTR) and closely related ultrafast spin-transfer studies.
Optical intersite spin transfer (OISTR) is an ultrafast, light-driven mechanism in which spin-polarized charge is redistributed between distinct atomic sites, elements, or magnetic sublattices during or immediately after femtosecond optical excitation, thereby modifying local magnetic moments on sub-100-fs timescales. In the formulations reported for transition-metal alloys, multilayers, Heusler compounds, and altermagnets, OISTR is characterized by coherent, optically induced intersite transitions that are initially spin conserving and whose efficiency is governed by the spin-resolved electronic structure, especially the availability of unoccupied final states near the Fermi energy [2008.12612, 2306.02793, 2305.16455]. In multicomponent ferromagnets such as FePt, CoPt, NiPt, FeNi alloys, and Co\(_2\)MnGa, the phenomenon is resolved as element- and site-specific moment changes tied to hybridized bands and pump-accessible optical transitions; in g-wave altermagnets such as CrSb, it appears as sublattice-selective demagnetization whose symmetry depends strongly on laser incidence direction and polarization [2008.12612, 2509.14991].

## 1. Definition and microscopic mechanism

OISTR is defined as a purely optically driven relocation of spin-polarized electrons between inequivalent sites or sublattices. In the Pt-containing 3d ferromagnets studied by Borchert, von Korff Schmising, Schick, Engel, Sharma, Shallcross, and Eisebitt, the relevant pathway is minority-spin transfer from Pt 5d states into unoccupied minority-spin 3d states of Fe, Co, or Ni, which increases the minority occupation at the ferromagnetic site and therefore reduces the local moment \(M_{\mathrm{FM}}(t)=\mu_B(n_\uparrow-n_\downarrow)\) because \(n_\downarrow\) increases while \(n_\uparrow\) is not concurrently reduced on the earliest timescales [2008.12612]. In Fe–Ni alloys, the reported mechanism is a minority-spin transfer from Ni-derived states below \(E_F\) into Fe-derived minority states above \(E_F\), producing an early-time anti-correlation between element-resolved responses [2306.02793, 2304.14957]. In Co\(_2\)MnGa, the dominant channel is Co minority 3d \(\rightarrow\) Mn minority 3d, whereas in g-wave CrSb OISTR is described as transfer between antiferromagnetically aligned Cr sublattices through optically allowed interband transitions between sublattice-resolved spin sectors [2305.16455, 2509.14991].

The microscopic driver is the time-dependent optical field acting on a hybridized, exchange-split band structure. The TDDFT description used in the Pt-based study writes the electronic dynamics through the time-dependent Kohn–Sham equation
\[
i\hbar \, \partial_t \Psi_{n\mathbf{k}}(\mathbf{r},t) \;=\; \hat{H}_{\mathrm{KS}}(t)\,\Psi_{n\mathbf{k}}(\mathbf{r},t),
\]
with \(\hat{H}_{\mathrm{KS}}(t)\) containing the time-dependent exchange–correlation potential, ionic potential, spin–orbit coupling, and light–matter coupling through a time-dependent vector potential \(\mathbf{A}(t)\) in velocity gauge [2008.12612]. A tight-binding current-operator picture gives the intersite current between sites \(i\) and \(j\) as
\[
\hat{J}_{ij} = \frac{ie}{\hbar} \sum_{\sigma} t_{ij} \, c_{i\sigma}^\dagger c_{j\sigma} \;+\; \mathrm{h.c.},
\]
which formalizes the notion that optical excitation populates spin-resolved final states on another site if those states are dipole-accessible through hybridization [2008.12612].

A central distinction in the literature is that OISTR is not identified with Elliott–Yafet spin-flip scattering, superdiffusive spin transport, or ordinary Stoner excitations. In Fe–Ni, OISTR is described as local to the alloy’s sublattices and spin conserving during the optically driven stage, in contrast to superdiffusive currents that export angular momentum spatially and Elliott–Yafet processes that rely on scattering and dominate on longer timescales [2306.02793]. In Co\(_2\)MnGa, the sub-100-fs dynamics are explicitly decomposed into three channels—same-site Co–Co transfer, intersite Co–Mn transfer, and SOC-mediated spin flips—showing that the measured magneto-optical asymmetry can contain signatures of several distinct microscopic processes at once [2305.16455].

## 2. Density of states, hybridization, and phase-space control

A recurrent result across the OISTR literature is that the availability of unoccupied final states near \(E_F\) controls the magnitude of the effect. In the Pt-based Fe, Co, and Ni systems, the relevant control parameter is the number of empty minority-spin 3d states above \(E_F\) within the pump-accessible window,
\[
N_{\downarrow}^{\mathrm{empty}} \;=\; \int_{E_F}^{E_F+\Delta E} \rho_{\downarrow}^{3d}(E)\, dE,
\]
where \(\rho_{\downarrow}^{3d}(E)\) is the minority-spin 3d density of states of the ferromagnet [2008.12612]. The reported ground-state DOS shows unoccupied minority-spin 3d regions extending roughly to \(\approx 2\) eV for FePt, \(\approx 1\) eV for CoPt, and \(\approx 0.5\) eV for NiPt. Since the pump photon energy is \(1.55\) eV, this implies the largest intersite phase space for Fe-based systems and progressively less for Co- and Ni-based systems [2008.12612].

The same paper presents a rate picture in which the optically induced intersite transition rate scales with the joint density of states,
\[
W_{i\to j}(\omega) = \frac{2\pi}{\hbar} \, |M_{ij}(\omega)|^2 \, D_{ij}(\omega),
\]
so that increased minority-spin DOS in the relevant window raises the demagnetization efficiency \(\eta\) [2008.12612]. This DOS-based interpretation is extended into materials design language: to enhance OISTR-driven magnetic control, one should increase the unoccupied minority-spin d-state phase space of the magnetic sublattice in the energy window addressed by the pump, for example through alloying with 5d elements such as Pt or by multilayer engineering that improves interfacial hybridization and band alignment [2008.12612].

In Fe–Ni alloys, the same logic is expressed through composition dependence rather than 3d–5d hybridization. The element-resolved EUV-MOKE study reports that the Ni increase in \(\mathrm{Re}\,\varepsilon_{xy}\) at \(E^* \approx 63.9\) eV is strongest in Fe\(_{50}\)Ni\(_{50}\), weaker in Fe\(_{19}\)Ni\(_{81}\), and smallest in pure Ni, matching the expectation that adding Fe introduces more Fe minority-spin final states above \(E_F\) and therefore enhances intersite transfer [2306.02793]. In Co\(_2\)MnGa, OISTR is attributed to a half-metallic electronic structure with many occupied Co minority-spin states below \(E_F\) and many unoccupied Mn minority-spin states above \(E_F\), combined with Co–Mn 3d hybridization near \(E_F\) [2305.16455]. In g-wave CrSb, the decisive factor is not simply total DOS but the presence of spin-compensated versus spin-uncompensated local DOS along momentum-space paths selected by the laser polarization; this introduces an explicitly \(k\)-selective generalization of the phase-space criterion [2509.14991].

A plausible implication is that OISTR is best viewed not as a single universal channel but as a family of optically activated, spin-preserving intersite transitions whose strength is set jointly by hybridization, optical matrix elements, and the spin-resolved availability of final states.

## 3. Experimental observables and verification strategies

The experimental literature emphasizes that OISTR must be inferred from element-, site-, or energy-resolved signatures rather than from bulk demagnetization alone. In the Pt-based ferromagnets, ultrafast magnetization dynamics were measured with two-color pump–probe magneto-optical Kerr effect (MOKE), using \(\lambda_{\mathrm{pump}}=800\) nm and \(\lambda_{\mathrm{probe}}=400\) nm, both with 39 fs FWHM, and a measured pump–probe cross-correlation of 55 fs [2008.12612]. The normalized magnetization was fit by a double-exponential response convolved with the 55 fs Gaussian cross-correlation,
\[
\frac{M(t)}{M_0} - 1 \;=\; G * \left[ H(t) \left( A\left(e^{-t/\tau_1} - 1\right) + B\left(1 - e^{-t/\tau_2}\right) \right) \right],
\]
and the demagnetization efficiency \(\Gamma\) was defined as the low-fluence slope of the demagnetization amplitude \(A\) versus absorbed fluence [2008.12612]. In that framework, OISTR is inferred from systematic material trends: the enhancement of early demagnetization in Pt-containing systems and the scaling of that enhancement with minority-spin DOS.

In Fe–Ni alloys, verification proved more subtle. The 2023 study on Fe\(_{50}\)Ni\(_{50}\), Fe\(_{19}\)Ni\(_{81}\), and pure Ni explicitly states that an increase in the Ni magneto-optical signal at a single energy is insufficient to verify OISTR, because the same feature appears in pure Ni, where no intersite Fe–Ni transfer is possible [2306.02793]. The work identifies two specific ambiguities. First, a transient increase at \(E^* \approx 63.9\) eV can arise from intra-Ni redistribution within the minority channel. Second, the EUV transverse MOKE asymmetry can invert sign near spectral zero crossings, and the observed transient depends sensitively on incidence angle; the same spectral region near \(64\) eV shows opposite transient changes for \(\theta=43.4^\circ\) and \(45.1^\circ\) solely due to geometry [2306.02793].

To overcome these ambiguities, the paper reconstructs the complex off-diagonal dielectric tensor component \(\varepsilon_{xy}(E,t)\) by global Fresnel-based fitting of spectrally resolved, multi-angle EUV T-MOKE. The measured asymmetry is
\[
A(E,\theta,t) = \frac{I_+(E,\theta,t)-I_-(E,\theta,t)}{I_+(E,\theta,t)+I_-(E,\theta,t)},
\]
and the central claim is that \(\varepsilon_{xy}(E,t)\), not raw asymmetry at a single energy, is the geometry-independent, element-specific quantity suitable for OISTR verification [2306.02793]. The verification protocol that emerges requires: reconstruction of \(\varepsilon_{xy}(E,t)\); element-resolved early-time anti-correlation; composition dependence consistent with acceptor-state availability; and direct comparison to TDDFT [2306.02793].

The Py substrate study uses a different diagnostic, defining an “OISTR trace” \(OT(t)\) from spin polarization in two narrow windows selected to follow Ni-derived states below \(E_F\) and Fe-derived states above \(E_F\),
\[
OT(t) = SP_{\mathrm{Ni}}(t) - SP_{\mathrm{Fe}}(t),
\]
with \(\Delta E=0.7\) eV windows at \(E=E_F-1.8\) eV and \(E=E_F+1.35\) eV [2304.14957]. There, OISTR is identified by the prompt increase of \(SP_{\mathrm{Ni}}\) and decrease of \(SP_{\mathrm{Fe}}\), whereas the suppression of that anti-correlation signals competition from interlayer spin export into Au [2304.14957].

## 4. Quantitative manifestations in representative material classes

The strongest systematic dataset on OISTR-driven demagnetization efficiency is the Fe/Co/Ni and Pt-based alloy/multilayer series. For elemental ferromagnets at absorbed fluence \(\approx 4\) mJ/cm\(^2\), Fe demagnetizes by \(\approx 10\%\), Co by \(\approx 18\%\), and Ni by \(\approx 80\%\), with extracted demagnetization efficiencies
\[
\Gamma_{\mathrm{Fe}} = 3.2 \pm 0.1,\quad
\Gamma_{\mathrm{Co}} = 7.3 \pm 0.7,\quad
\Gamma_{\mathrm{Ni}} = 23.7 \pm 1.7
\]
in \%/(mJ/cm\(^2\)), which is a monotonic increase from Fe to Co to Ni [2008.12612]. Pt-containing multicomponent systems demagnetize much more efficiently:
\[
\Gamma_{\mathrm{Fe|Pt}} = 40.2 \pm 3.5,\quad
\Gamma_{\mathrm{Co|Pt}} = 71.0 \pm 2.3,\quad
\Gamma_{\mathrm{Ni|Pt}} = 120.3 \pm 8.7,
\]
while ordered alloys give
\[
\Gamma_{\mathrm{FePt}} = 58.1 \pm 2.3,\quad
\Gamma_{\mathrm{CoPt}} = 66.5 \pm 8.2
\]
in the same units [2008.12612]. The gain relative to the pure element decreases across the series, with multilayer ratios \(\Gamma(\mathrm{FM|Pt})/\Gamma(\mathrm{FM})\) of approximately \(12.6\) for Fe, \(9.7\) for Co, and \(5.1\) for Ni, which the paper interprets as a direct consequence of decreasing empty minority-spin phase space as the d band fills [2008.12612].

In Fe–Ni alloys, the most specific quantitative OISTR signatures are temporal and element resolved rather than bulk-efficiency based. The reconstructed \(\varepsilon_{xy}(E,t)\) yields demagnetization constants and amplitudes for Fe\(_{50}\)Ni\(_{50}\):
\[
\tau_{\mathrm{Fe}} = 215 \pm 6 \ \mathrm{fs},\quad
\tau_{\mathrm{Ni}} = 184 \pm 19 \ \mathrm{fs},
\]
\[
\Delta A_m(\mathrm{Fe}) = 0.286 \pm 0.003,\quad
\Delta A_m(\mathrm{Ni}) = 0.234 \pm 0.008,
\]
and for Fe\(_{19}\)Ni\(_{81}\):
\[
\tau_{\mathrm{Fe}} = 147 \pm 4 \ \mathrm{fs},\quad
\tau_{\mathrm{Ni}} = 153 \pm 3 \ \mathrm{fs},
\]
\[
\Delta A_m(\mathrm{Fe}) = 0.525 \pm 0.004,\quad
\Delta A_m(\mathrm{Ni}) = 0.498 \pm 0.004
\]
[2306.02793]. The onset delay \(t_0(\mathrm{Ni})-t_0(\mathrm{Fe})\) is reported as \(12 \pm 3\) fs in Fe\(_{19}\)Ni\(_{81}\) and \(95 \pm 7\) fs in Fe\(_{50}\)Ni\(_{50}\), with the larger delay tracking stronger OISTR [2306.02793].

In Co\(_2\)MnGa, TDDFT identifies population changes at a 70 fs snapshot showing depletions in Co minority 3d below \(E_F\) and increases in Mn minority 3d above \(E_F\). The resulting net moment dynamics are reported as an increase of \(M_{\mathrm{Co}}\) by up to \(2.5\%\) and a decrease of \(M_{\mathrm{Mn}}\) by \(\sim 2\%\) within the pump [2305.16455]. The EUV TMOKE signal depends strongly on probe energy: the transient enhancement is \(\sim 5\%\) at the Co peak \(60.4\) eV, \(\sim 14\%\) at \(62.5\) eV, greater than \(100\%\) near the \(59.3\) eV zero crossing, and greater than \(2000\%\) at \(71.5\) eV, where the ground-state asymmetry baseline is very small [2305.16455]. These values are interpreted not as direct moment changes alone but as the combined spectroscopic consequence of same-site transfer, intersite transfer, and spin flips.

In CrSb, OISTR manifests as anisotropic sublattice demagnetization rather than conventional ferromagnetic quenching. Under normal incidence along \([0001]\), the two Cr sublattices demagnetize symmetrically and net magnetization remains negligible, but under off-normal incidence with \(\phi=\pm56^\circ\) and \(\theta=0^\circ\) or \(60^\circ\), asymmetric demagnetization begins after \(\approx 15\) fs and a transient ferrimagnetic-like state emerges [2509.14991]. The net magnetization from TDDFT reaches on the order of several \(10^{-2}\,\mu_B\), with the angle map spanning roughly \(-0.06\) to \(+0.06\,\mu_B\) [2509.14991].

## 5. Competition with spin flips, interlayer transport, and other channels

A consistent theme in the literature is that OISTR rarely acts in isolation. In Pt-containing 3d ferromagnets, the earliest demagnetization appears within the experimental 55 fs resolution and continues over a few hundred femtoseconds, where coherent electronic processes dominate; later-time superdiffusive transport and magnonic processes are considered unable to account for the observed correlation with minority-spin DOS on the earliest timescales [2008.12612]. The same work also shows, in CoPt, that with SOC turned off OISTR alone yields a loss in Co moment nearly compensated by a gain in Pt moment, whereas with SOC on the demagnetization is amplified because Pt majority and minority states can both feed Co minority states through SOC-enabled channels [2008.12612]. Thus OISTR and SOC are not exclusive mechanisms; they can co-occur and reinforce each other.

The Py/Au study examines a different competition: intralayer OISTR versus interlayer spin transfer into a substrate. In Py/MgO, where interlayer transport is suppressed, the OISTR fingerprint is strong: \(SP_{\mathrm{Ni}}\) increases while \(SP_{\mathrm{Fe}}\) decreases as soon as the pump arrives, \(OT(t)\) peaks at about \(200\) fs, and the imbalance decays within the next \(\approx 250\) fs [2304.14957]. Replacing MgO with Au opens interlayer pathways that drain or refill the very carrier populations OISTR creates. For Py/Au(10 nm)/MgO, the \(OT\) magnitude is clearly reduced; for Py/Au(100 nm)\), the \(OT\) vanishes entirely, indicating that the Au substrate acts as an efficient spin sink and interlayer spin transfer dominates the earliest dynamics when not transiently blocked by substrate state filling [2304.14957]. The same paper introduces the notion of “state blocking” in Au: in thin Au, optical absorption and confinement of deposited energy near the interface reduce the available final-state phase space for interlayer transport over the pump duration and subsequent several hundred femtoseconds, permitting a residual OISTR signature [2304.14957].

Co\(_2\)MnGa adds a further layer of complexity because same-site Co–Co transfer strongly affects the energy dependence of the TMOKE asymmetry. The paper reports that early times \(0\)–\(50\) fs are dominated by SOC-mediated spin flips near \(E_F\), mid-pump \(50\)–\(80\) fs by OISTR and same-site Co–Co transfer, and post-pump by relaxation with renewed spin-flip dominance [2305.16455]. At the Co peak \(60.4\) eV, the enhancement grows with fluence only up to \(\sim 1.0\) mJ/cm\(^2\) absorbed and then diminishes and shifts earlier because demagnetization channels outcompete OISTR experimentally, even though TDDFT predicts monotonic enhancement with fluence [2305.16455]. This divergence is explicitly attributed to channels absent from the one-unit-cell TDDFT treatment, including magnons, electron–phonon scattering, and superdiffusive spin currents [2305.16455].

A plausible generalization from these studies is that OISTR is best understood as the earliest coherent redistribution channel within a broader hierarchy of ultrafast magnetic processes. Whether it dominates the observable response depends on the competition posed by SOC-mediated spin flips, interlayer transport, and later incoherent scattering.

## 6. Controversies, limitations, and current interpretive landscape

The existence of OISTR as a useful explanatory framework is strongly supported in some systems and actively debated in others. The Pt-based ferromagnet study reports a clear DOS-controlled scaling of demagnetization efficiency, supported by TDDFT that reproduces both the monotonic Fe \(\rightarrow\) Co \(\rightarrow\) Ni trend in pure elements and the enhanced efficiencies of Pt-containing systems [2008.12612]. The Fe–Ni verification study similarly argues that, once geometry artifacts and single-energy ambiguities are removed, OISTR can be identified unambiguously in Fe\(_{50}\)Ni\(_{50}\) and, more weakly, in Fe\(_{19}\)Ni\(_{81}\), through geometry-independent \(\varepsilon_{xy}(E,t)\), composition dependence, and TDDFT-predicted Fe/Ni anti-correlation [2306.02793].

However, the 2025 fluence-dependent FeNi study challenges the OISTR interpretation of delayed Ni demagnetization. It states that TDDFT-based OISTR for FeNi predicts mirrored sub-50-fs changes in Fe and Ni moments, including a transient Ni moment increase and stronger OISTR signatures at higher fluence, because the intersite transfer rate \(k_{\mathrm{Ni\to Fe}}(F,\omega)\) should grow strongly with pump field amplitude [2503.08811]. Experimentally, the opposite fluence trend is reported: the Fe–Ni delay \(\Delta t(F)\) decreases monotonically from \(\sim 72\) fs at \(2\) mJ/cm\(^2\) to below \(15\) fs at \(14\) mJ/cm\(^2\) in T-MOKE, while transmission MCD gives \(42 \pm 21\) fs at \(\sim 6\) mJ/cm\(^2\) and \(11 \pm 11\) fs at \(\sim 20\) mJ/cm\(^2\) [2503.08811]. No transient increase of the Ni moment is observed at any fluence, and Fe demagnetization saturates at approximately \(5 \pm 1\%\) at the onset of Ni demagnetization across fluences [2503.08811]. On that basis, the paper concludes that, under its conditions, OISTR is not supported as the origin of Fe–Ni timing differences and that an exchange-mediated inhomogeneous magnon generation scenario with spin-wave instability better explains the data [2503.08811].

This disagreement does not invalidate OISTR as a mechanism but narrows the conditions under which it can be cleanly verified. The verification paper itself cautions that a Ni increase at a single energy is insufficient evidence and that geometry-dependent asymmetry sign changes can mimic or obscure transfer signatures [2306.02793]. The FeNi controversy therefore concerns not only the underlying physics but also the standards of experimental proof. One side argues that multi-angle \(\varepsilon_{xy}\) reconstruction together with TDDFT yields unambiguous intersite transfer evidence [2306.02793]; the other argues that fluence dependence and the absence of a transient Ni moment increase are inconsistent with the expected OISTR phenomenology in Fe\(_{0.5}\)Ni\(_{0.5}\) [2503.08811].

Theoretical limitations are also repeatedly noted. In the Pt-based work, ordered L1\(_0\) alloys are used as theoretical proxies for experimental solid-solution alloys and multilayers [2008.12612]. In Co\(_2\)MnGa, the one-unit-cell TDDFT neglects magnons, electron–phonon scattering, and superdiffusive transport, which likely explains why theory overestimates fluence-dependent OISTR at high fluence [2305.16455]. In CrSb, the reported dynamics are purely electronic TDDFT results on tens-of-femtoseconds timescales; phonon-mediated relaxation lies outside the modeled window [2509.14991]. These caveats suggest that OISTR is most rigorously established in the earliest coherent interval after excitation, whereas later-time behavior generally requires additional channels.

Taken together, the literature presents OISTR as a robust ultrafast mechanism for intersite redistribution of spin-polarized carriers, a quantitatively predictive concept when the spin-resolved DOS and optical matrix elements are favorable, and a phenomenon whose experimental identification demands element-specific, geometry-aware, and often theory-assisted analysis. Its broader significance lies in providing a materials-design principle for femtosecond spin control: engineer hybridization and spin-resolved phase space so that the pump couples occupied donor states to unoccupied acceptor states on the desired sublattice, while controlling competing channels such as interlayer spin export, SOC-driven spin flips, and incoherent scattering [2008.12612, 2304.14957, 2305.16455, 2509.14991].

Source: https://www.emergentmind.com/topics/optical-intersite-spin-transfer-oistr