Skip to content

Development of a modular optically detected magnetic resonance setup for optical experiments in a variable temperature insert

FieldValue
JournalReview of Scientific Instruments
Published2026-07-01
AuthorsAnh Tong, Andreas Bauer, Markus Kleinhans, J. S. Schilling, C. H. Back
DOI10.1063/5.0337322

An optically detected magnetic resonance (ODMR) setup delivered nitrogen-vacancy (NV) magnetometry through an approximately 190 cm free-space path inside a commercial helium bath cryostat. The optical head stayed outside, while the sample remained inside a 30 mm variable temperature insert with a 1.6 K to 300 K operating range. The authors recorded spectra from 1.6 K to 240 K. Field tests and a SrRuO3 transition confirmed usable signal and alignment without significant cryostat modification.

A modular route into a commercial cryostat

Section titled “A modular route into a commercial cryostat”

The target system was an Oxford Instruments recondensing IntegraAC cryostat with a 16 T superconducting magnet. Its narrow VTI bore and long optical access made stable alignment difficult. The authors adapted the optical experiment rather than the cryostat itself.

The setup combined an external optical head, a custom sample stick, and a rail-guided positioning platform. The head handled excitation and fluorescence detection, while the stick carried microwave and optical components.

A room-temperature support frame allowed pre-alignment and testing before installation. Its spring-loaded platform also decoupled mechanical stress from the sample stick. Most reinstallation and cooldown cycles required no optical adjustment. Rare corrections used less than 1° of dichroic-mirror rotation. The platform could move sideways for access to the helium filling port.

The optical head used a fiber-coupled 532 nm laser and two mirrors on kinematic gimbals. A 567 nm longpass dichroic reflected the excitation beam downward through the cryostat. Returned fluorescence passed through two 650 nm longpass filters and a neutral density filter before a silicon avalanche photodiode. The neutral density filter kept bright samples within the detector’s linear range.

An Olympus 10x objective had an NA of 0.25 and a 21 mm working distance. It focused excitation onto the sample and collected NV fluorescence. At that NA, the objective captured about 1.5% of isotropic emission. For a 5 mm beam and an 18 mm focal length, the estimated excitation diameter was 5 to 6 µm. The test diamond measured 3 × 3 × 0.5 mm³. Its surface was {100} ± 3°, with a homogeneous NV concentration of about 4.5 ppm. Applied fields were perpendicular to that surface.

The lower holder integrated the objective, a Cernox sensor, a custom coplanar waveguide, the sample, and a non-magnetic linear z-piezo actuator. Three 160 cm stainless steel rods formed the triangular sample stick. Baffles reduced radiative heat transfer and stray light.

Measurement sequence and spectral analysis

Section titled “Measurement sequence and spectral analysis”

A continuous-wave 532 nm laser typically supplied 100 mW. The temperature and field spectra used 4 W microwave source power. The microwave was switched on and off for intervals of 16 µs. A 1 µs detection window sampled each interval, and 10,000 repetitions produced each averaged point.

The authors formed an ODMR spectrum from the ratio of fluorescence recorded with and without microwave driving. This isolated spin-dependent contrast and suppressed background. It also normalized laser and detector fluctuations. A Python script fitted two Lorentzians and extracted resonance frequency, FWHM, and contrast.

At 241 K under nominal zero field, the fitted separation was 4.37 MHz. This corresponded to an effective residual field of 0.078 mT. The authors associated the apparent splitting with lattice strain and a small residual field. They listed magnet remanence, the Earth’s field, and stray fields as possible contributions. This background splitting set the baseline for later field analysis.

The temperature series used the 4.5 ppm NV chip at nominal zero field. Microwave source power was 4 W, and laser power was 100 mW. The resonance-center shift agreed with the analytical model of Doherty et al. down to about 20 K. Their reference temperature was 5 K, while the present data used 1.6 K. Below 20 K, deviations became apparent.

The authors suggested local laser and microwave heating could raise the sample above the thermometer reading. This becomes more likely in laminar helium flow. They optimized helium flow and microwave power to reduce heating while retaining sufficient signal-to-noise.

ODMR contrast decreased as temperature fell, reached a minimum near 30 K, and recovered slightly below that point. The authors related the intermediate-temperature decrease to thermally activated orbital and spin mixing. They related the low-temperature recovery to reduced phonon-mediated mixing. The trend agreed qualitatively with prior measurements reported by Ernst et al.

At 200 K, the authors varied the applied field from nominal zero to 26 mT along the (100) surface normal. Microwave and laser powers were 4.0 W and 100 mW. The four NV orientations in this geometry lie at 54.7° to the surface normal. Their parallel and perpendicular field factors were 0.577 and 0.816.

The measured parallel field came from the resonance separation using a gyromagnetic ratio of 28 GHz T⁻¹. It agreed with the expected projection of the applied field. The reconstructed total field also matched the applied magnitude. As field increased, both fitted linewidths grew and both contrasts fell.

The authors attributed the broadening to ensemble inhomogeneity and partly to transverse-field spin mixing. The mixing reduced spin polarization and fluorescence contrast. Device analysis must therefore account for all four NV families and their local strain.

A second-order treatment of the perpendicular field reproduced the asymmetric displacement of both resonances. The orientation-only and strain-corrected curves nearly overlapped, showing that strain had a weak effect on the overall field shift. Despite the magnet’s 16 T maximum, reliable ODMR was effectively limited to the mT regime. Microwave bandwidth and transverse spin-state mixing caused this limit.

The sample measured approximately 1.7 × 0.68 × 0.16 mm and sat between the microwave stripline and the diamond chip. Its [110] direction was parallel to the stripline. Its [001] direction was aligned with the out-of-plane magnetic field.

Under a 5 mT bias field, ODMR spectra changed across the SrRuO3 ferromagnetic transition. The paper reports spectra at 153 K below the transition and 236 K above it. An additional resonance splitting appeared below the transition. The authors attributed this change to the onset of ferromagnetic order.

A Quantum Design MPMS 3 measurement used the same 5 mT bias and a field along pseudo-cubic [001]. Its magnetization rose sharply on cooling through approximately 164 K. The matching transition in both methods supported the ODMR platform for cryogenic studies of correlated magnetic materials.

The authors estimated sensitivity from a T2* of 0.5 µs in their commercial diamond chip. Their lower-bound field sensitivity was about 230 nT/√Hz under moderate conditions. This corresponds to about 230 nT over a 1 s measurement time. At sensor distances of 50 to 100 nm, they estimated detectable moments of about 10⁻²² to 10⁻²¹ A m².

The estimate reaches moment scales 10 to 11 orders of magnitude below the cited VSM limit of about 10⁻¹¹ A m². The advantage comes from sensor proximity, not superior absolute field resolution. A 20 × 10 × 0.1 µm SrRuO3 lamella carried about 3 × 10⁻¹² A m² at low temperature. At 100 nm, its estimated stray field was about 50 to 100 mT, more than five orders above the ODMR threshold.

Several limits remain. Reliable temperature control reached about 20 K, although spectra were recorded at lower temperatures. Cryocooler vibration caused no observable linewidth, contrast, or spectral-quality change at the reported resolution. The authors warn that vibration may matter for scanning magnetometry requiring sub-micron stability. Compatibility with diamond anvil cells is a design goal, not a demonstrated measurement. Further work must optimize microwave delivery and optical coupling for that geometry.

The paper’s sensing element is a 0.5 mm thick, {100}-oriented single-crystal diamond chip containing about 4.5 ppm NV centers for 532 nm ODMR. The diamond materials catalog lists MPCVD-grown single-crystal diamond from 0.1 to 500 micrometers thick, while 6CCVD also lists metallization with Au, Pt, Pd, Ti, W, or Cu. The engineering support page can support a review of those options against the chip geometry and cryogenic feedthrough interfaces required by the coplanar-waveguide design.

View Original Abstract

We developed an optically detected magnetic resonance setup designed for compatibility with a widely used, commercially available helium bath cryostat equipped with a variable temperature insert. The optical path extends nearly 2 m, spanning the full length of the cryostat insert, enabling excitation of the nitrogen-vacancy (NV) centers and detection of the resulting fluorescence from outside the cryostat. The setup preserves optical alignment and beam quality along this extended path, allowing integration into existing cryogenic systems without significant modifications. We demonstrate the setup’s performance by measuring the temperature dependence of the resonance signal and its behavior under small applied magnetic fields, as well as the magnetic transition of a SrRuO3 sample, thereby showcasing the feasibility of NV magnetometry on a sample in constrained cryogenic environments.

  1. 2017 - High Sensitivity Magnetometers [Crossref]