Volume fraction hides where the new phase forms
Standard ultrafast X-ray diffraction tracks a phase transition through the integrated intensity of each phase's Bragg peak. That ratio gives the volume fraction of the ferromagnetic phase, VFM(t), but it averages over the whole probed volume. A sample that converts uniformly and one that converts in a few narrow columns can produce the same curve.
The paper states this limitation directly: following only VFM(t), as earlier FeRh work did, provides no insight into transient phase heterogeneity. The new measurement is designed to recover the missing spatial information rather than to sharpen the same averaged number.
A strain pulse marks each phase separately
The method, called ultrafast X-ray sonography, combines hard-X-ray diffraction with a picosecond acoustic strain pulse travelling through the film. When the pulse enters or leaves a layer, the compressive and expansive parts inside that layer become unbalanced, so the layer's average strain shifts and its Bragg peak moves.
Because the antiferromagnetic and ferromagnetic phases have distinct Bragg peaks, each phase's peak shift reports that phase's own strain response. The delay between the compression and expansion maxima of a phase, Δtphase, gives its thickness through dphase = Δtphase/vs, where vs is the longitudinal sound velocity. The delay between the two phases' compression maxima locates the depth of the boundary between them.
Combining the relative thickness Dphase(t) with the measured volume fraction Vphase(t) yields the in-plane coverage via Vphase(t) = Dphase(t)·Aphase(t). This is how a depth-averaged intensity ratio becomes a three-dimensional description of where each phase sits.
In FeRh the excitation also launches the probe
The conceptual scheme uses a separate photoacoustic transducer to create the strain pulse. The FeRh experiment needs none: the same optical excitation that drives the transition also launches the strain pulse that probes it. The authors note this makes the approach largely independent of sample preparation.
The measurement used 44 nm FeRh(001) films grown on MgO(001), capped with 2 nm Pt and buffered by 8 nm W, excited at room temperature (300 K) with 800 nm pulses at incident fluences from 3.9 to 11.7 mJ cm−2. The reported sonogram was taken at 7.7 mJ cm−2. Lattice probing used 11.2 keV X-ray pulses at the MID instrument of EuXFEL, with the sample placed between the antiferromagnetic Bragg peak at 21.76° and the ferromagnetic peak at 22.04°.
Ferromagnetic domains grow as ~30 nm columns
The extracted phase-specific strain responses show the ferromagnetic phase nucleating at the surface in narrow, vertically extended columnar domains with a diameter of approximately 30 nm. The authors describe this as directly visualising a process previously obscured in spatially averaging measurements.
The 30 nm figure is not read off an image. It comes from matching an extra broadening of 0.06 Å−1 in the ferromagnetic Bragg peak to a model in which neighbouring nucleation sites reverse each other's in-plane expansion. Smaller domains would leave a larger average in-plane strain; larger domains a smaller one. Accounting for the simplified five-site building block and the assumption that all extra broadening comes from broken in-plane symmetry, the authors assign a 10 nm uncertainty to this in-plane determination.
Fast nucleation fits stochastic site-by-site growth
The ferromagnetic volume fraction rises as a single exponential, VFM(t) = H(t)·V*FM·(1 − e^(−t/τ)), with τ = 8 ps under this non-equilibrium optical pathway. The authors interpret this form as stochastic nucleation at independent sites over an Arrhenius energy barrier, and contrast the 8 ps with about 50 ps for near-equilibrium heating.
That kinetic reading and the spatial reading are separate results. The exponential describes how much ferromagnetic volume appears over time; the sonography describes where it appears. The paper's claim is that both are needed, because the same averaged rise is compatible with very different spatial arrangements.
Directional heterogeneity reconciles earlier FeRh results
The authors present the columnar, surface-nucleated picture as common ground for interpreting the diverse experimental observations on FeRh, and they emphasise directional heterogeneity as a control on the kinetics of first-order transitions. The discussion also notes that the domain size matches the mosaic crystal blocks of the FeRh layer, which the authors take to point to structural granularity as relevant to how long-range order emerges.
That link between domain size and mosaic blocks is the authors' interpretation, not a measured causal test. The experiment establishes the domain geometry and the nucleation kinetics; it does not isolate which structural feature sets the 30 nm scale.
The probe needs distinguishable diffraction signatures
The method applies to phase transitions with structural changes as long as the coexisting phases have distinguishable diffraction signatures. Its effective temporal resolution is set by the strain pulse transit time, that is, sound speed times phase thickness. The authors argue this matters mainly for pure symmetry-changing transitions, since the appearance of a structural phase through volume change is itself limited by the speed of sound.
The authors also distinguish sonography from real-space imaging, which they say is typically restricted to systems where the dynamics are not stochastic, for example because defects pin the transition. Sonography is presented as sensitive to both coherent growth and stochastic nucleation, and the authors propose double-pulse schemes with a band-gap transducer to mark heterogeneity at a chosen pump-pump delay.
Read the 30 nm domain size as a bounded estimate
The result worth carrying forward is not a single number but a pairing: a surface-nucleated, columnar ferromagnetic phase about 30 nm wide, forming with an 8 ps exponential rise. The width carries a stated 10 nm uncertainty and depends on a five-nucleation-site model; the 8 ps timescale belongs to the non-equilibrium optical pathway at the reported fluences, not to near-equilibrium heating, where the paper cites about 50 ps.
For anyone comparing FeRh reports, the paper's priority is that in-plane and out-of-plane heterogeneity must be separated before volume fractions are interpreted. A curve of VFM(t) alone cannot distinguish a uniform front from columns, and the sonography measurement is what supplies that distinction.
Sources used in this answer
Ultrafast X-ray sonography reveals the spatial heterogeneity of the laser-induced magneto-structural phase transition in FeRh
The ferromagnetic phase in laser-excited FeRh nucleates at the surface as narrow columnar domains of approximately 30 nm diameter, resolved by combining ultrafast hard-X-ray diffraction with a propagating strain pulse.
