In a recent paper in Proceedings of the National Academy of Sciences (PNAS) Barry et. al.  demonstrate sensing of a single neuron action potential (AP) using a nitrogen-vacancy (NV) sensor on a diamond chip. Diamonds are an attractive option for a sensor material due to their inherent scalability and biocompatibility. This mean that it is in theory possible to monitor multiple neurons in close proximity whilst being able to discern between them, as the authors themselves say:
“With further development, we expect micrometer-scale magnetic imaging of a variety of neuronal phenomena.”
First though it is useful to clarify two things:
What is the action potential of a neuron?
What is a NV center?
To answer the first question a gif from Wikipedia is very helpful (see figure 2). The cell starts at equilibrium with a potential of around -70 mV. When the cell receives an impulse ion channels on the membrane separating the charges open leading to a rapid increase in potential (equivalent to shorting out a capacitor). The increase of potential above zero is related to the the ions present in the cell environment and kind of spoils my capacitor analogy… Once a positive voltage within the cell has been reached ion channels close and ion pumps within the cell restore the potential to its resting value of -70 mV.
In the research paper scientists however did not directly measure the potential but rather the magnetic field induced by the time-varying charge distribution with the cell. Equation 1 describing the dynamics describes the dynamics where B is the magnetic field 𝚽 - the voltage across the cell wall and S is a scaling factor dependant on the geometry and local environment.
The reason why is to do with the structure of their essential measurement device - the nitrogen vacancy center in diamond (neat segway huh?). The name gives the game away, a NV center is a type of defect in diamond where one of the carbon atoms is replaced with a nitrogen atom while another adjacant atom is simply missing. An excess electron is also present in the defect making the defect negatively charged.
The reason why these defects are used as magnetic field probes is that that their fluorescence is dependant on the the spin state the defect is in. If we illuminate a diamond with these defects for a certain amount of time we will transfer all the population to the m=0 state and get maximum red fluorescence. However if we apply microwaves at just the right frequency (such that the energy matches that of either m=-1 or m=+1 spin transition) we observe a dip in the fluorescence. Figure 3 shows the energy diagram of the ground state of an NV center and optically detectable magnetic resonance (ODMR) spectrum.
It is possible to probe the magnetic field because the energy splitting between different spin states is dependent on the magnetic field due to the Zeeman interaction (equation 2).
So after all this the main idea is simple, we have a mechanism that generates a magnetic field and we have a method to measure it with high magnetic and spatial resolution. And that is what they got (see figure 4). A suction was used to stimulate the nerve and the diamond is placed underneath the worm. The authors report the worm surviving without any noticable change in behaviour even after prolonged experiments. The experiments were conducted on three different species.
However everything is not so simple (which is why the work is published in such a high impact factor journal). The magnetic field generated by the axon is on the order of nanotesla (for comparison the earths magnetic field is on the order of microtesla - a thousand times stronger). The theoretical maximal magnetic field sensitivity of a diamond probe working in the continuous regime (there exist sophisticated pulsed measurement procedures which are past the scope of this article) is 3.1 pT/(Hz)½ . However after taking into account noise from electronic components and other setup-related sources it grows to 13.8 pT/(Hz)½ . At a temporal resolution of 2 𝛍s this yields a magnetic field resolution of around 4 nT and a signal-to-noise ratio of around SNR = 1. In other words - even after extensive optimization of the diamond and experimental setup (see supplementary information of ) time-averaging is necessary to measure single-fire events reliably
The authors were able to monitor the magnetic field generated by a live organism for an extended period of time without killing it however the sensitivity of a continuous wave (laser and microwaves always on, no pulse sequences used) measurement is not high enough to reliably measure single-fire events and time-averaging over multiple impulses is necessary.
 Marblestone, Adam H. et al. “Physical Principles for Scalable Neural Recording.” Frontiers in Computational Neuroscience 7 (2013): 137. PMC. Web. 5 Dec. 2016.
 Barry, J. F., Turner, M. J., Schloss, J. M., Glenn, D. R., Song, Y., Lukin, M. D., … Walsworth, R. L. (2016). Optical magnetic detection of single-neuron action potentials using quantum defects in diamond. Proceedings of the National Academy of Sciences, 201601513. https://doi.org/10.1073/pnas.1601513113