Microwave-free vector magnetometry and crystal orientation determination with Nitrogen-Vacancy centers using Bayesian inference

This paper presents a microwave-free vector magnetometry framework using Nitrogen-Vacancy centers in diamond that leverages Bayesian inference and cross-relaxation resonances to simultaneously determine magnetic field vectors and crystal orientations from photoluminescence maps, thereby enabling compact, alignment-free quantum sensing without the heating and interference issues associated with traditional microwave techniques.

The Gist

Imagine you have a tiny, invisible compass hidden inside a diamond. This compass isn't made of metal; it's made of a specific defect in the diamond's crystal structure called a Nitrogen-Vacancy (NV) center. Scientists have long known these centers are amazing at sensing magnetic fields, but usually, to read them, you have to blast the diamond with microwaves. Think of this like trying to listen to a whisper in a room while someone is constantly shouting into a megaphone. The microwaves heat things up and create interference, which makes it hard to use these sensors in delicate or compact situations.

This paper introduces a clever new way to listen to that whisper without the megaphone.

The Problem: The "Microwave Headache"

Traditionally, to figure out the direction and strength of a magnetic field using these diamond sensors, researchers use a technique called ODMR. This involves tuning the sensor with microwaves to see when it "resonates" (like a guitar string vibrating).

• The Issue: Microwaves are messy. They heat up equipment, require bulky wires, and can disturb the very thing you are trying to measure.
• The Old "No-Microwave" Attempts: Some previous methods tried to avoid microwaves by looking at how the diamond glows (photoluminescence) when different internal compasses interact. However, these methods were very picky. They required the external magnetic field to be perfectly aligned with the diamond's crystal axes. It was like trying to open a lock only if you held the key at a precise 90-degree angle; if you were off by a degree, it didn't work.

The Solution: The "Bayesian Detective"

The authors propose a new method that is microwave-free and alignment-free. They use a mathematical tool called Bayesian inference.

Here is the analogy:
Imagine you are in a dark room with a complex, multi-faceted crystal chandelier. You can't see the crystal, but you can see how the light reflects off it when you move a flashlight around.

1. The Setup: You shine a light on the diamond and rotate it while slowly changing the strength of a magnetic field.
2. The Clue: As you do this, the diamond's glow (photoluminescence) dips at specific moments. These dips happen because the different "compasses" inside the diamond (which point in four different directions) suddenly start talking to each other. This conversation is called cross-relaxation.
3. The Map: If you plot these dips on a 2D map (one axis is the rotation angle, the other is the magnetic field strength), you get a unique pattern of valleys and ridges.
4. The Detective Work: Instead of guessing, the team uses a "Bayesian Detective" (a computer algorithm). This detective doesn't just look for one peak; it looks at the entire map of the glow. It asks: "Given this entire pattern of dips, what is the most likely direction the diamond is facing, and what is the magnetic field doing?"

How It Works (The "Magic" of the Math)

The paper explains that the diamond has a specific symmetry (like a tetrahedron, or a pyramid with four sides). Because of this, there are multiple ways the diamond could be oriented that would produce the exact same glow pattern.

• The Challenge: A simple computer might get confused and say, "It's either facing North or South," and pick one randomly.
• The Fix: The Bayesian method is smart. It doesn't force a single answer. Instead, it produces a probability map. It says, "There is a 50% chance it's facing North and a 50% chance it's facing South." It acknowledges the ambiguity naturally, rather than forcing a wrong answer.

What They Actually Did

The researchers didn't just theorize this; they built it in a lab.

1. Orientation Test: They took a diamond with a random, unknown orientation. They shone light on it, rotated the magnetic field, and recorded the glow. The algorithm successfully figured out exactly how the diamond was sitting in space, identifying two possible "mirror image" orientations that fit the data perfectly.
2. Magnetometry Test: Once they knew how the diamond was oriented, they used the same method to measure an unknown magnetic field. They rotated the diamond and changed the field, and the algorithm successfully reconstructed the full 3D vector (direction and strength) of the magnetic field.

Why This Matters (According to the Paper)

• No Microwaves: It removes the need for heating or complex microwave wiring.
• No Perfect Alignment: You don't need to carefully line up the magnetic field with the diamond. You can just spin the diamond (or the field) and let the math figure out the rest.
• Robustness: It works even with noisy data and handles the confusing "mirror image" possibilities of the diamond's symmetry gracefully.

In short, the paper presents a new "smart camera" for magnetic fields. Instead of needing a perfectly aligned, microwave-blasted setup, it takes a picture of how a diamond glows while it spins, and uses advanced math to reverse-engineer both the diamond's position and the magnetic field's strength and direction. This paves the way for smaller, simpler, and more practical magnetic sensors.

Technical Summary

Technical Summary: Microwave-free Vector Magnetometry and Crystal Orientation Determination with Nitrogen-Vacancy Centers using Bayesian Inference

Problem Statement
Nitrogen-vacancy (NV) centers in diamond are established as high-sensitivity quantum sensors for magnetometry, thermometry, and electric field detection. Conventional NV magnetometry typically relies on Optically Detected Magnetic Resonance (ODMR), which requires microwave (MW) fields to manipulate spin states. While robust, MW delivery introduces practical limitations, including environmental perturbation, heating, and the need for galvanic connections that complicate integration, particularly under cryogenic conditions.

Microwave-free alternatives exist, such as those exploiting ground-state level anticrossing or magnetic field-induced mixing of spin sublevels. However, these methods often require stringent alignment of the external magnetic field with specific crystallographic axes or high-resolution optical instrumentation to map photoluminescence (PL) patterns. Furthermore, a critical bottleneck in vector magnetometry is the necessity of knowing the precise crystallographic orientation of the NV centers. Traditional calibration involves ODMR spectroscopy under controlled fields, while MW-free orientation methods are often limited by imaging precision and dipole modeling. Recent work utilizing NV-NV cross-relaxation has demonstrated sensitivity but has been constrained by protocols requiring prior crystallographic knowledge and precise field alignment to unambiguously assign resonances.

Methodology
The authors propose a general framework for microwave-free vector magnetometry and crystal orientation determination that leverages Bayesian inference to analyze 2D photoluminescence (PL) maps. The core of the method relies on the cross-relaxation phenomenon: when spin transitions of inequivalent NV orientations become degenerate, flip-flop processes are enhanced, leading to a measurable drop in PL intensity.

1. Physical Model: The system utilizes a tunable bias magnetic field ($B_{bias}$) applied along a fixed axis (laboratory $z$-axis) and a static external field ($B_{ext}$). The diamond sample is rotated around this axis (or equivalently, the transverse field is rotated). The total magnetic field in the sample frame is determined by the rotation angle $\phi$ and the bias strength.
2. Resonance Conditions: Cross-relaxation occurs on specific planar surfaces in magnetic field space defined by linear relations between field components (e.g., $B^s_x = \pm B^s_y$). These conditions manifest as sharp dips in the PL signal. The model accounts for the discrete degeneracies inherent to the NV tetrahedral symmetry ($O$ group), where multiple physically distinct orientations yield identical PL patterns.
3. Analytical Forward Model: Instead of repeatedly diagonalizing the Hamiltonian, the authors derive a closed-form analytical expression for the expected PL signal. This model incorporates the resonance conditions as Lorentzian-like dips, parameterized by the external field vector, the sample orientation matrix, linewidth, and contrast weights.
4. Bayesian Inference: The inverse problem—extracting the magnetic field vector and crystal orientation from the PL map—is solved using Bayesian estimation.
   • Posterior Distribution: The method computes the full posterior probability distribution $P(\theta|y)$, where $\theta$ represents the unknown parameters (field components, orientation angles) and $y$ is the experimental PL data.
   • Handling Degeneracy: Unlike peak-fitting methods that force a single solution, the Bayesian approach naturally captures discrete degeneracies as distinct peaks in the posterior distribution, quantifying uncertainty and ambiguity.
   • Symmetry Reduction: To ensure computational efficiency, the parameter space for orientation is restricted to a minimal, symmetry-inequivalent domain by exploiting the tetrahedral symmetry of the NV centers, reducing the search space from 24 equivalent configurations to a manageable subset.

Key Contributions

• Alignment-Free Framework: The work introduces a protocol that does not require prior knowledge of the crystal orientation or precise alignment of the external field with crystallographic axes. Both the magnetic field vector and the NV orientation are extracted simultaneously from the data.
• Analytical Efficiency: By deriving an analytical model for cross-relaxation resonances, the authors avoid the computational bottleneck of repeated Hamiltonian diagonalization, enabling fast and scalable Bayesian inference.
• Robustness to Degeneracy: The framework explicitly handles the discrete symmetries of the NV system, providing full posterior distributions that reveal multiple valid solutions (e.g., inversion-related orientations) rather than discarding them or selecting an arbitrary best fit.
• Experimental Demonstration: The authors experimentally validate the method using a wide-field fluorescence microscope and a diamond ensemble. They successfully reconstructed the orientation of a diamond crystal with an unknown alignment and subsequently used this calibrated orientation to reconstruct an unknown external magnetic field vector.

Results

• Orientation Determination: In a proof-of-concept experiment, the team determined the orientation of a diamond crystal relative to the laboratory frame. The Bayesian analysis of the experimental PL map yielded two high-probability regions corresponding to inversion-related orientations, consistent with the system's symmetry. The reconstructed parameters ($\alpha, \beta, \zeta$) reproduced the experimental PL pattern with high fidelity.
• Vector Magnetometry: Using the previously determined orientation, the authors reconstructed an unknown external magnetic field. The marginal posterior distributions for the field components ($b_z, b_\perp, \phi_0$) showed well-localized peaks. The reconstructed values were compatible with independent ODMR measurements.
• Symmetry Analysis: Simulations demonstrated that when the rotation axis aligns with high-symmetry crystal directions (e.g., [100], [110], [111]), the PL maps exhibit rotational and mirror symmetries that lead to ambiguities in field reconstruction. However, by choosing an arbitrary rotation axis (breaking symmetry), unique reconstruction is achieved.

Significance
The paper establishes a general route toward compact, alignment-free NV magnetometers suitable for practical sensing applications where conventional ODMR techniques are impractical due to heating or integration constraints. By leveraging Bayesian inference on cross-relaxation features, the method removes the need for stringent field alignment and prior crystallographic knowledge, offering a robust solution for vector magnetometry and orientation determination in arbitrary field configurations. The work broadens the scope of NV-based sensing by providing a framework that naturally quantifies uncertainties and handles the intrinsic discrete degeneracies of the NV symmetry.