Skip to content

Nanoscale NV-Center Magnetometry of a Synthetic Three-Dimensional Spin Texture

FieldValue
JournalThe Journal of Physical Chemistry C
Published2026-03-10
AuthorsRicardo Javier Peña Román, Sandip Maity, Fabian Samad, Dinesh Pinto, S. Josephy
DOI10.1021/acs.jpcc.6c00518

Ambient nitrogen-vacancy scanning probe microscopy produced the first quantitative vector-field magnetometry measurements for the multilayered synthetic antiferromagnet [(Co/Pt)5/Co/Ru]3/(Co/Pt)6. Maximum static stray fields reached approximately ±8.0 mT at antiferromagnetic domain boundaries. The maps used 3.5 nm pixels and a 36 to 76 nm probe distance.

The stack was deposited at room temperature by dc magnetron sputtering. The base pressure was 8 × 10⁻⁶ Pa, and Ar sputter gas was held at 0.4 Pa. A silicon substrate carried 100 nm thermally grown SiO₂. The stack above it included 1.5 nm Ta, a 20 nm Pt seed layer, and a 3 nm Pt oxidation cap.

The magnetic stack contained four Co/Pt ferromagnetic blocks separated by 0.75 nm Ru spacers. Its full form was [[Co(0.55 nm)/Pt(0.70 nm)]5/Co(0.55 nm)/Ru(0.75 nm)]3/[Co(0.55 nm)/Pt(0.70 nm)]6. Each ferromagnetic block was 6.8 nm thick. SQUID-VSM measured the saturation magnetization as (833 ± 83) kA/m.

NV-SPM used 515 nm optical initialization and 650 to 850 nm photoluminescence readout under ambient conditions. The authors used noncontact amplitude modulation for all images. Quantitative ODMR and dual-iso-B imaging used a commercial scanning microscope with an arbitrary-direction electromagnet.

Single-NV probes had nominal implantation depths of 10 nm and diamond orientations of (100), (110), or (111). The estimated DC field sensitivity was about 2 μT/Hz. MFM served as a reference measurement, using a 15 nm tip-sample distance in lift mode. Unlike NV-SPM, MFM applies a tip field of a few tens of mT that can reversibly change or irreversibly reorient sample moments.

MFM maps of the remanent state showed micrometer-sized antiferromagnetic domains separated by ferromagnetic stripes about 100 nm wide. The stripe contrast had an average periodicity of about 300 nm.

A zero-field NV-PL map from the (100) probe showed bright antiferromagnetic domains and about 40% quenching along ferromagnetic boundary stripes. Local ODMR on an antiferromagnetic domain had symmetric Zeeman splitting near 2.87 GHz and about 20% contrast. On a stripe, the splitting was asymmetric and contrast fell below 10%.

The asymmetric spectrum and several-mT transverse field led the authors to assign the quenching to spin mixing with this probe. Some stripe positions produced fields too strong for ODMR.

An external 20 mT field at 103° to the NV axis produced reversible, MFM-like contrast. The authors’ model assumed ±5 mT stripe fields, giving transverse totals of 24 and 15 mT. It predicted 23% relative contrast at 20 mT and 7% at 60 mT.

Contrast changes therefore arose from compensation or competition between external and sample fields, rather than a magnetic tip interacting with the sample.

Quantitative field mapping with a (111) probe

Section titled “Quantitative field mapping with a (111) probe”

A (111) probe placed the NV axis at 180° to the laboratory vertical, nearly collinear with the sample magnetization. The authors used symmetric ODMR to rule out spin mixing for this geometry. The projected field was 0.6 mT on an antiferromagnetic domain and 4.0 mT on a ferromagnetic stripe.

Dual iso-B maps resolved contours at -4.6 and +6.1 mT. They also found a zero-field line between domains and zero-field areas within an antiferromagnetic domain. The authors interpreted the zero-field line as the out-of-plane field reversal and used it as a domain-wall position marker.

ODMR vector maps found maximum static fields of about ±8.0 mT at antiferromagnetic domain boundaries. Across the antiferromagnetic domains, in-plane and out-of-plane components ranged from hundreds of μT to 1.0 mT. Fourier reverse propagation reconstructed the two in-plane components from the measured projection.

Granular contrast within antiferromagnetic domains indicated incomplete magnetization compensation. The authors linked this granularity to nonuniform film thickness and local surface variation in the sputtered multilayer. The small residual field can also arise because the NV center sits at unequal distances from the four ferromagnetic blocks.

The zero-field line defines the domain-wall center only between similar domains. At a boundary between ferromagnetic and antiferromagnetic regions, the extended stray field shifts that line toward the antiferromagnetic domain. The shift reaches several tens of nanometers at 75 nm height.

Three-dimensional texture and micromagnetic agreement

Section titled “Three-dimensional texture and micromagnetic agreement”

MuMax3 modeled four ferromagnetic blocks, each 6.75 nm thick, separated by 0.75 nm spacers. The 2.048 μm × 2.048 μm simulation volume extended 75 nm above the sample. The simulation used 4 × 4 × 0.75 nm³ cells.

A 2 nm lateral test changed the overall texture little but increased memory use and computation time. The model used measured Ms = 833 kA/m, exchange stiffness 8.0 pJ/m, and interlayer coupling of -0.7 mJ/m². It also used perpendicular anisotropy of 0.6 MJ/m³ and a 32 nm average grain size. Second-order anisotropy and interfacial DMI were omitted.

Simulation produced alternating domain-wall shifts in adjacent ferromagnetic blocks. These shifts create ferromagnetic cores within an undulated three-dimensional texture. The measured shifts ranged from about 0 to 40 nm. Earlier work reported a nearly constant 20 nm shift for comparable X = 6 and N = 4 structures.

Simulated stray fields reached about ±12 mT, while experiments reached about ±8.0 mT. The spatial patterns agreed, including domain-boundary modulation and granular noncompensation. The higher simulated amplitude suggests that the magnetic parameters and NV distance still require optimization.

Prior models linked the shifts to dipolar and antiferromagnetic exchange competition. Other work found that domain-wall pinning also affects stripe stability. The present data support nonuniform shifts and nanoscale ferromagnetic cores, but do not directly establish the wall type.

All-optical T₁ measurements initialized the NV state with a 3 μs laser pulse. After a variable waiting interval, a second pulse read out the state. The baseline relaxation time before approaching the sample was (1370 ± 142) μs.

Over an antiferromagnetic domain, T₁ fell to (90 ± 9) μs. Over a ferromagnetic stripe, it fell to (14 ± 1) μs. The authors attributed these reductions to GHz magnetic noise near the NV transition at 2.87 GHz, most likely from thermal magnons.

The authors associated the shorter stripe value with confined thermal magnonic modes or stronger noise matching the NV resonance. Similar T₁ changes with (100), (110), and (111) probes indicated perpendicular noise components relative to each NV axis.

The (100) stripe quench remained dominated by spin mixing. The (111) stripe quench had no local ODMR evidence for spin mixing, so noise was the dominant cause. Reduced laser power did not improve the contrast, which the authors linked to the ferromagnetic-core field splitting the NV mₛ = ±1 states.

Engineering implications and open questions

Section titled “Engineering implications and open questions”

The probe orientation controls whether NV-SPM can measure a thick synthetic antiferromagnet quantitatively. A collinear (111) axis avoided the transverse fields that mixed the NV spin states. An off-axis (100) probe produced useful qualitative contrast but required compensation models for field interpretation.

Probe height remains a design uncertainty. Simulations over 36 to 75 nm shifted the inferred wall position by about 3 nm, corresponding to about 8% of the total height variation. Height calibration and local magnetometry therefore affect wall-position estimates.

The direct wall-type determination remains unresolved. The thin-sample stray-field formula is invalid because the multilayer thickness is comparable with the sensor distance. The authors proposed refined micromagnetic simulations or Lorentz transmission electron microscopy for that test.

For synthetic-antiferromagnet design, the measurements identify two local variables that conventional imaging cannot separate: transverse static fields and GHz noise. Both can quench the same NV photoluminescence signal. A calibrated vector map should therefore be paired with T₁ relaxometry before assigning a stripe signal to spin-wave behavior.

The T₁ data identify strong local noise, but they do not directly measure the spin-wave dispersion. Thermal magnons remain the authors’ most likely source.

The paper’s sensor is an NV-active, crystallographically oriented single-crystal diamond scanning tip with shallow NV implantation at a nominal depth of 10 nm. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal diamond, but it does not specify NV implantation or finished-tip geometry. The probe requirements would need separate qualification through the linked engineering support.

View Original Abstract

Multilayered synthetic antiferromagnets (SAFs) are artificial three-dimensional (3D) architectures engineered to create novel, complex, and stable spin textures. Noninvasive and quantitative nanoscale magnetic imaging of the two-dimensional stray field profile at the sample surface is essential for understanding the fundamental properties of the spin-structure and being able to tailor them to achieve new functionalities. However, the deterministic detection of spin textures and their quantitative characterization at the nanoscale remain challenging. Here, we use nitrogen-vacancy scanning probe microscopy (NV-SPM) under ambient conditions to perform the first quantitative vector-field magnetometry measurements in the multilayered SAF [(Co/Pt)5/Co/Ru]3/(Co/Pt)6. We investigate the static and dynamic nanoscale properties of antiferromagnetic domains with boundaries hosting “one-dimensional” ferromagnetic stripes with ∼100 nm of width and periodic modulation of the magnetization. By employing NV-SPM measurements in different imaging modes and involving NV-probes with various crystallographic orientations, we demonstrate distinct fingerprints emerging from GHz-range spin noise and constant stray fields on the order of several mT. This provides quantitative insights into the structure of domains and domain walls, as well as, into magnetic noise associated with thermal spin-waves. Our work opens up new opportunities for quantitative vector-field magnetometry of modern magnetic materials with tailored 3D spin textures and stray field profiles, and potentially novel spin-wave dispersions─in a quantitative and noninvasive manner, with exceptional magnetic sensitivity and nanometer scale spatial resolution.