Paper: Camera-limits for diamond magnetometry

While doing experimental physics it is not uncommon that technical challenges take up a big portion of the time spent in the lab. Choosing the right equipment, setting it up and fine-tuning it to your application can become a time-sink. Planning your experiments are just as much part of research as the actual execution is, but it tends to be overlooked. Too often practical details and knowledge gathered about the equipment are lost since they are rarely touched upon in conventional physics publications. Fortunately, there are journals that do not just acknowledge the relevance of preserving such information but even focus on them. In the March issue of Review of Scientific Instruments, which (as the name suggests) is one such journal, work done at bigQ has been featured.

Working with wide-field magnetometry based on color centers in diamond, one relies on detecting minute changes in the weak fluorescence signal emitted from the Nitrogen-Vacancy centers (NV centers). For this, the detection apparatus is obviously of great importance and to do wide-field imaging the detector has to be a camera. Various options are available on the market and an in-depth understanding of their capabilities and limits is of course essential in making a choice.

The paper Camera-limits for wide-field magnetic resonance imaging of nitrogen-vacancy spin sensors by Adam M. Wojciechowski et al. briefly introduces NV magnetometry using optically detected magnetic resonance (ODMR). Sensitivity can be understood as the minimal detectable change in the magnetic field. This means that we want a low number for the sensitivity. To improve the sensitivity one has to increase the signal contrast and/or the signal to noise ratio. If line broadening occurs, it will however degrade the sensitivity.

Source: Adapted from paper
Example of optically detected magnetic resonance (ODMR) recorded with lock-in detection. The depth of the resonance dips governs the contrast, the signal to noise ratio is mainly about the noise in the background level and the linewidth/line-broadening is the frequency span of the resonances.

The paper then goes into detail with how different camera specifications affect the achievable sensitivity. The limits of the camera will limit the entirety of the setup regardless of all other components. Here is a brief summary of the findings for 5 different camera specifications:

Full well capacity

Source: https://commons.wikimedia.org/wiki/File:Well,_Prague_Satalice.jpg

The full well capacity is how much light the sensor can handle before it gets overexposed and results in distortion of the image. So, even though we generally want as much signal as possible, it is necessary that the camera does not get overwhelmed. Since this limits the magnitude of the signal, it also puts an upper bound on the signal to noise ratio.

Camera framerate

Source: https://commons.wikimedia.org/wiki/File:USDOT_highway_sign_bicycle_symbol_-_black.svg

This is a matter of how fast the camera can operate, namely how many photoelectrons can be collected per unit time. The camera speed will be limited by the processing time of transferring the signal, but it does not limit the sensitivity itself as long as there is enough light for illumination within each frame.

Sensor resolution

Most laboratory cameras offer a resolution in the megapixel range, with typical pixel dimensions being between 1-10 µm. Larger pixels provide a bigger full well capacity but at the cost of a lower spatial resolution. Another important factor for the sensor is the digital resolution of the analog-to-digital converters in the camera. Having enough bits is crucial to maintain a high dynamic range, which means that you are able to detect very small differences in the signal. The only problem with having more bits is that it then takes longer to transfer them out of the camera, which in turn limits the overall detection speed.

Quantum efficiency and fill factor

Source: https://commons.wikimedia.org/wiki/File:CMOS_Image_Sensor_Big_and_Small.jpg

In reality, any detector converts only a fraction of all the incoming photons into photoelectrons. The quantum efficiency is a measure of the certainty that an incoming photon will result in an electrical signal. The fill factor of a camera refers to what portion of the overall area of a pixel is actually photosensitive. For most cameras, both these values are quite good and in practice, they do not limit the sensitivity.

In-pixel lock-in detection

A new type of camera called the time-of-flight camera has lock-in detection built into each pixel. It requires a demodulation of the signal and timing of the triggering so that the noise in the signal can be reduced by automatically subtracting a background measurement. Using a time-of-flight camera one can improve the magnetic sensitivity by a factor of 1/√(4*contrast).

Source: https://www.flickr.com/photos/140988606@N08/28457522300

These are the basic limitations in the different camera features. On top of this, it is possible to do post-processing to increase sensitivity at the cost of something else – for example by binning over either time or space.

These are just the limiting factors for the cameras, but in the paper you can find a comparison of 5 different laboratory cameras. The quick summary is that some of the ‘normal’ cameras do better when it comes to single frames, but if you want to go fast it seems that you have to use a time-of-flight camera. Even in an unoptimized configuration such a camera allowed us to measure magnetic fields with µT sensitivity and video frame rates.

Leave a comment

Your email address will not be published. Required fields are marked *