HOW IT WORKS

In a nutshell

Scanning NV

We use a single defect in diamond, an NV center, as an atomic size quantum sensor. By scanning it in close proximity over a sample surface, we can quantitatively map magnetic fields and more with high sensitivity and better than 30 nm resolution. All of this works in ambient conditions in a desktop size instrument.

NV centers explained in depth

Nitrogen Vacancy centers

A single defect, also known as color center, in diamond is at the heart of our technology. A Nitrogen Vacancy center is created by removing two adjacent carbon atoms from the crystal lattice and placing one nitrogen at one of the sites. This "artificial atom" has several unique properties that make it interesting for a wide range of applications:


  • Its energy levels create a room temperature quantum bit with an easily accessible energy spacing in the GHz range.
  • The levels can be readout via an optical signal.
  • The gyromagnetic ratio of 28 GHz/T makes it very sensitive to magnetic fields
  • Lifetimes in the ms range and coherence times in the hundreds of microseconds (deep NVs).
  • NVs can be created artificially in high purity synthetic diamond.


NVs are a robust and popular platform for quantum sensing, as well as few qubit quantum computing, single photon sources and more.

Readout

The nitrogen vacancy center can be thought of as a two level system, with an energy spacing in the microwave range. In addition, it will emit red light when excited with green light. State readout is based on the fact that the excited state couples to a "dark state", reducing the fluorescence rate compared to the ground state.


Not only is green laser light used to read out the state, but it also initializes the system to the ground state.


By sweeping the microwave frequency and measuring the fluorescence at each frequency point, we can optically detect the resonance.

Zeeman Splitting

The fundamental B field detection relies on the splitting of the excited state. By measuring the frequency shift between the two excited states, the projection of the magnetic field onto the NV axis can be measured.

Sensitivity

Magnetic sensitivity strongly depends on the protocol used. To start, let's look at cw-ODMR, measuring at a single point frequency point. The shot noise limited sensitivity is then given by: 

Where C_cw is contrast, is the linewidth and Δν is the photon rate.

In cw-ODMR, both contrast and countrate can be increased at the expense of linewidth. Optimal sensitivity is typically achieved at low linewidths rather than high contrast or count rates.

What values can you actually expect? Our top of the line sensors have a contrast of better than 25% at a linewidth of 10 MHz and a countrate of 450 kcts/s, leading to a theoritical sensitivity of better than 1.6 μT/√Hz in cw-ODMR.

In practice, there are a few more complications:

  1. Measuring at a single frequency only works for very small magnetic field variations as well as constant contrast, linewidth and countrate.
  2. Countrates drop in close proximity to the sample (less waveguiding, cavity effects and more) and vary over the sample.
  3. Contrast and linewidth can change in close proximity of the sample.
  4. The stated formula implies a shot noise limited setup.

As a result of 2 & 3, scanning NV often measures ODMR spectra to obtain the full information (see Measurement Modes). In this case, the actual sensitivity, defined as the standard deviation of the determined field in a given time (including both integration time and software overhead) can deviate significantly from the theoretical formula. For example, the QSM can achieve the shot noise limit in an optimized setting at a known magnetic field. In a standard setting used for scanning it will still be within a factor of two to three of this limit.

Pulsed ODMR can significantly decrease the linewidth and thus improve the sensitivity by a factor of approximately seven. Whether this resuls in faster scanning depends on the dynamic range of the sample (overall frequency shift compared to the linewidth), since the sensitivity also depends on the frequency window size when taking specta. In a simple explanation, all the information is contained in the resonance. Using a frequency window that is large compared to the linewidth adds measurement time without adding information, which decreases sensitivity. Decreasing the linewidth but maintaining the window size because of the dynamic range may thus not add sensitivity. The QSM offers resonance tracking to mitigate this issue.

Other pulsed methods, such as AC magnetometry and Ramsey, can also increase sensitivity by almost an order of magnitude. They may not be applicable (AC magnetometry) or require significant overhead to keep tuned in an actual scanning measurement (Ramsey).


Frequently asked questions

A scanning NV magnetometer is a next-generation scanning probe microscope that utilizes a so-called nitrogen-vacancy center (NV center) as an atomic size magnetic field sensor. The NV center is a lattice defect in diamond whose properties we manipulate with microwave frequencies and whose state we determine by optical fluorescence. Because NV centers are essentially qubits, even at room temperature, the scanning NV magnetometer can exploit quantum metrology techniques to achieve very high magnetic field sensitivity. Scanning NV magnetometers are a major breakthrough in high-resolution and passive magnetic analysis of surfaces.

In a simple picture, the NV has a magnetic orientation which we can flip using a specific microwave frequency. If we manage to flip it, the fluorescence rate is reduced. The amount of energy needed to flip the spin (frequency of the microwave) depends on the external magnetic field and its orientation. For increasing field we need more energy to flip the parallel orientation and less to flip the antiparallel one. The frequency difference between these two cases only depends on the magnetic field and the gyromagnetic ratio, which is a natural constant, creating a sensor that does not need to be calibrated.

We fabricate scanning probes where the NV center is located at the apex and ~10 nm below the surface. Using atomic force microscopy (AFM) feedback techniques, we then scan the NV over the surface of interest at a height of 20–50 nm. The result is a quantitative map of the surface magnetic field with high sensitivity and resolution.

We can use NVs’ sensitivity to magnetic fields to also measure magnetic surface noise or current and current densities. In addition, the NV is also sensitive to temperature and electric field, making it a very versatile surface characterization probe.

Magnetic sensitivity strongly depends on the protocol used. In cw-ODMR, our top-of-the-line sensors have a contrast of 25% at a linewidth of 10 MHz and a count rate of 450 kcts/s, leading to a theoretical sensitivity of better than 1.6 μT/√Hz.

Will you be able to achieve this value in your setup? This depends on several factors:

  • How efficiently does your setup collect light? How good is the confocal rejection?
  • Is your setup and data acquisition shot-noise limited?
  • How much does close proximity to your sample reduce count rate and contrast?

Pulsed protocols, such as pulsed ODMR or AC magnetometry, can further improve the sensitivity by a factor of seven or more.

Spatial resolution, defined as how close two identical features can be placed and still distinguished, is determined by the NV-surface distance, typically called stand-off or flying height. While we implant for a nominal NV depth of 10 nm, the statistical nature of the implantation as well as the AFM setup lead to a stand-off between 30 nm and 70 nm, resulting in a similar resolution. The QSM offers an enhanced spatial resolution mode that consistently measures at lower stand-off.

The QSM can measure ODMR spectra in scanning mode at 200 pixels/s, for an overall acquisition time for a 1350×760 pixel image (1 MP) of less than 2 hours. This is possible for samples that have a dynamic range greater than 200 μT, using our top-of-the-line tips, at a noise floor of approximately 20 μT.

For smaller shifts that require a lower noise floor, measurement time increases accordingly. Similarly, samples that quench contrast and/or count rates increase measurement time.