X-PEEM: Imaging the Nanoscale Magnetic World
[Note: This page uses MathJax to render TeX equations]
Prerequisite Reading: If you aren't familiar with X-ray Magnetic Circular/Linear Dichroism, check out the XMCD & XMLD Deep Dive first!
1. From Spectroscopy to Microscopy
Standard X-ray Absorption Spectroscopy (XAS) is an incredible tool, but it has one major limitation: it averages the signal over the entire footprint of the X-ray beam (which can be tens to hundreds of microns wide). If your sample has magnetic domains that are only 100 nm across, XAS will just measure a macroscopic average of zero.
Enter X-ray Photoemission Electron Microscopy (X-PEEM). Rather than just counting the total number of absorbed X-rays, X-PEEM uses electron optics to preserve the precise spatial origin of the absorption events, achieving a spatial resolution down to 20–50 nm.
The Secondary Electron Cascade
How do we "see" X-ray absorption? When an incident X-ray is absorbed by a core electron, that electron is excited to the valence band, leaving behind a highly unstable core hole. Within femtoseconds, an upper-level electron drops down to fill the hole. This process releases energy, which is often transferred to a third electron that gets ejected from the atom—an Auger electron.
As this high-energy Auger electron travels toward the surface, it scatters off other electrons, creating a massive avalanche of low-energy secondary electrons. The crucial physics here is that the Total Electron Yield (TEY) is directly proportional to the initial X-ray absorption cross-section.
In an X-PEEM instrument (like MAXPEEM at MAX IV), a massive extraction voltage (\( \sim 10-20 \text{ kV} \)) is applied between the sample and the objective lens. This violently rips the secondary electrons away from the surface and accelerates them through a stack of electron lenses, ultimately projecting a highly magnified spatial map of the electron yield onto a detector.
2. XMCD-PEEM: Imaging Ferromagnetic Domains
In a ferromagnet, the X-ray absorption depends on the alignment between the X-ray's circular polarization helicity (\( \boldsymbol{\sigma} \)) and the local magnetization vector (\( \mathbf{M} \)). But a raw PEEM image contains a massive amount of "garbage" contrast: topography (bumps and divots), work function variations, and chemical inhomogeneity.
To isolate the purely magnetic signal, we use the Asymmetry Calculation. We take two identical images: one with right-circularly polarized light (\( I_+ \)) and one with left-circularly polarized light (\( I_- \)). We then calculate the asymmetry map pixel-by-pixel: $$ A(x,y) = \frac{I_+(x,y) - I_-(x,y)}{I_+(x,y) + I_-(x,y)} $$
Because topography and chemistry are independent of the X-ray polarization, they appear identically in both \( I_+ \) and \( I_- \). When we subtract the images, those artifacts perfectly cancel out. When we divide by the sum, we normalize the intensity, leaving a pristine map where the pixel intensity is directly proportional to \( \mathbf{M} \cdot \boldsymbol{\sigma} \). Magnetic domains pointing parallel to the X-rays appear bright (white), antiparallel domains appear dark (black), and perpendicular domains appear neutral (gray).
3. XMLD-PEEM: Imaging the Invisible Antiferromagnet
Antiferromagnets (AFMs) are notoriously difficult to image. Because their adjacent spins point in opposite directions, the net macroscopic magnetization is strictly zero. They produce no stray magnetic fields, rendering techniques like Magnetic Force Microscopy (MFM) entirely blind.
However, we know that X-ray Magnetic Linear Dichroism (XMLD) scales with \( 3\cos^2\theta - 1 \), meaning it depends on the axis of the spins (the Néel vector \( \mathbf{n} \)), not their absolute direction. This makes XMLD-PEEM one of the only techniques capable of resolving antiferromagnetic domains at the nanoscale.
To map AFM domains, we typically use linearly polarized X-rays and exploit the multiplet splitting of the \( L_3 \) edge. For example, in NiO, the \( L_3 \) edge is split into two distinct peaks (often denoted \( L_3^A \) and \( L_3^B \)). The XMLD contrast reverses sign between these two peaks.
By taking an image at the first peak energy (\( I_{E1} \)) and a second image at the second peak energy (\( I_{E2} \)), we can perform the same asymmetry calculation: $$ A_{XMLD}(x,y) = \frac{I_{E1}(x,y) - I_{E2}(x,y)}{I_{E1}(x,y) + I_{E2}(x,y)} $$ This yields a map of the antiferromagnetic twin domains, cleanly separating domains whose Néel vectors lie along different crystallographic axes.