Ultrafast magnetization induced by linearly polarized pulses is widespread in nonmagnetic semiconductors

Technical Summary: Ultrafast Magnetization Induced by Linearly Polarized Pulses in Nonmagnetic Semiconductors

Problem Statement
The ultrafast control of spin degrees of freedom is a critical requirement for quantum information, high-speed spintronics, and petahertz-scale information processing. While ultrafast demagnetization in magnetic metals (e.g., Ni) is well-established, the light-induced ultrafast magnetization of non-magnetic semiconductors remains a significant challenge. Specifically, switching a material from a "bit-zero" (non-magnetic) to a "bit-one" (finite magnetization) state within hundreds of femtoseconds using linearly polarized light has been theoretically proposed but rarely observed. Unlike circularly polarized light, which induces magnetization via the inverse Faraday effect and spin-orbit coupling, linearly polarized light carries no net angular momentum. Consequently, the mechanism for inducing magnetization with linear pulses is less understood, and clear guidelines for identifying suitable candidate materials have been lacking. Previous theoretical work suggested that systems with narrow band edges, high densities of states, or strong exchange interactions could exhibit a Stoner-like instability under photoexcitation, but this had only been demonstrated in a very limited number of theoretical cases.

Methodology
The authors employed a high-throughput first-principles screening workflow to systematically identify non-magnetic semiconductors capable of developing photoinduced magnetization.

• Database: The study utilized the Materials Cloud three-dimensional structure (MC3D) database of experimentally known inorganic crystals.
• Initial Filtering: The search was restricted to transition-metal-containing compounds with primitive unit cells of at most 14 atoms. Materials were selected if they were non-magnetic in their ground state and exhibited semiconducting character with a PBEsol Kohn-Sham band gap between 0.1 and 3 eV. This yielded an initial set of approximately 1,700 materials.
• Photoexcitation Simulation: To mimic ultrafast photoexcitation, the authors employed constrained density-functional theory (cDFT). They imposed a quasi-equilibrium distribution of electrons and holes by removing electrons from the valence band and injecting them into the conduction band, creating a fixed carrier density of 0.2 electrons per unit cell.
• Magnetic Instability Check: The electronic structure was recalculated self-consistently under these constrained occupations, allowing spontaneous spin polarization to emerge if energetically favorable. Spin-orbit coupling (SOC) was excluded at this stage to isolate exchange-driven instabilities.
• Gap Correction: Recognizing that PBEsol underestimates band gaps, the authors applied a hierarchical correction strategy for the promising candidates: using Modified Becke–Johnson (MBJ) gaps where available, all-electron HSE06 values otherwise, and machine-learning models trained on MBJ data for the remainder.
• Classification: Candidates were classified based on their local coordination motifs (octahedral, tetrahedral, planar, pyramidal, linear) and chemical composition to uncover structure-property relationships.

Key Results
The screening identified nearly 440 non-magnetic semiconductors that develop finite magnetization under carrier excitation, indicating proximity to an exchange instability.

• Widespread Occurrence: The phenomenon is not limited to isolated compounds but is widespread across diverse material families. Approximately 280 materials possess band gaps in the experimentally relevant optical window (1.5–3.5 eV) and exhibit absolute magnetic moments greater than 1 $\mu_B$ per photoexcited electron.
• Magnetic Order Diversity: The induced magnetic order is not restricted to a single configuration. The study identified ferromagnetic (FM), antiferromagnetic (AFM), and ferrimagnetic (FR) states.
  • High-Moment Group: Materials with moments approaching 2 $\mu_B$ per electron are dominated by fluorides (e.g., $ABF_6$, $A_2BF_6$) and specific low-dimensional oxides.
  • Lower-Moment Group: Spinels, delafossites, noble-metal oxides, and metal carbides typically cluster around 1.2 $\mu_B$ per electron.
• Structure-Property Motifs: The magnitude and type of magnetic order are governed by the crystal-field environment and the orbital character of the band edges:
  • Octahedral Motifs: $d^0$ systems often show ferrimagnetic order via $p \to d$ charge transfer. $d^{10}$ systems exhibit ferromagnetism driven by valence-band holes. Low-spin $d^6$ systems (especially fluorides) show strong ferromagnetic responses ($\approx 2 \mu_B$) due to localized $d \to d$ excitations within the same metal-ligand unit.
  • Tetrahedral Motifs: $d^0$ systems generally show ferrimagnetic order, while specific CrO$_4$ tetrahedral systems can exhibit large AFM moments.
  • Planar Motifs: $d^8$ square-planar systems predominantly stabilize ferromagnetic order.
  • Pyramidal Motifs: $d^0$ systems based on V, Mo, and W (e.g., VO$_5$ pyramids) often yield large AFM local moments exceeding 2 $\mu_B$.
  • Linear Motifs: $d^{10}$ compounds with linear coordination generally show ferromagnetic order driven by valence-band holes.
• Chemical Trends: Fluorides exhibit the highest average induced moments due to the localized nature of F 2p orbitals, which narrows band dispersions and enhances exchange splitting. Among transition metals, 3d elements (V, Co, Ni) generally show stronger magnetization than 4d or 5d elements due to stronger orbital localization and correlation effects.

Representative Case Studies
The paper details three representative mechanisms:

1. CrO$_3$ (Tetrahedral $d^0$): Photoexcitation creates electrons on Cr and holes on O sublattices, leading to opposite spin polarizations and an AFM state.
2. CuAlO$_2$ (Linear $d^{10}$): Magnetization is driven by holes in the narrow Cu 3d valence band, with conduction electrons following the exchange field, resulting in a ferromagnetic state.
3. Cs$_2$PdF$_6$ (Octahedral $d^6$): Both valence and conduction band edges arise from the same PdF$_6$ unit. Photoexcited electrons and holes both contribute to exchange splitting with the same spin sign, yielding a large ferromagnetic moment ($\approx 2 \mu_B$).

Significance and Claims
The authors argue that ultrafast magnetization induced by linearly polarized femtosecond pulses is a widespread occurrence in non-magnetic semiconductors, rather than a rare exception. The study provides:

• A systematic design strategy that shifts the focus from seeking strong spin-orbit coupling to identifying semiconductors with band-edge electronic structures favorable for exchange enhancement under photoexcitation.
• Intuitive guidance for materials selection based on coordination motifs and chemical trends (e.g., favoring fluorides and specific transition metals).
• A materials-level framework for discovering and designing semiconductors for optical control of magnetic properties.

The paper identifies numerous promising candidates for experimental verification, including vanadium-based compounds (e.g., VOPO$_4$, CaV$_2$O$_6$) and functional materials like NaNbO$_3$ and Cs$_2$TiCl$_6$. The authors acknowledge limitations, noting that spin-orbit coupling was not included in the screening (potentially affecting 5d compounds) and that calculated band gaps are estimates. However, the results suggest that the carrier-driven Stoner instability is compatible with realistic optical excitation energies, opening new avenues for experimental verification and application in ultrafast spintronics.