News & Views | Published:

Applied physics

Optical sensing gets exceptional

Nature volume 548, pages 161162 (10 August 2017) | Download Citation

Conventional sensors based on devices called optical resonators have fundamentally limited sensitivity. Careful engineering has been used to overcome this constraint, opening the door to ultraprecise sensing. See Letters p.187 & p.192

Light-confining structures called optical resonators make ideal sensors for a range of phenomena, including mechanical vibrations and changes in temperature, as well as for nanoparticles and biomolecules such as viruses and DNA1. However, the performance of conventional sensors based on optical resonators is fundamentally limited — halving the strength of the perturbation being sensed halves the ability of the device to sense it. On pages 187 and 192, respectively, Hodaei et al.2 and Chen et al.3 demonstrate a mechanism that avoids this limitation, paving the way for sensors that have unprecedented sensitivity to tiny changes in their environments.

Optical-resonator sensors can be understood by considering a simple example in which particles are detected using a ring-shaped or a toroidal resonator. Light travels around the resonator, and if the distance it moves in one cycle is equal to an integer multiple of the light's wavelength in the material, the optical power builds up — a phenomenon known as resonance. Modes of light are produced that have particular resonance frequencies, and the optical spectrum of the light consists of sharp peaks at these frequencies. The fewer photons that are lost in the resonator, the more times the light can travel around the device, and the sharper are the peaks in the optical spectrum.

If a particle approaches such a resonator, the optical path of the light changes and the resonance frequencies shift. The particle's presence can therefore be inferred by observing this shift in the optical spectrum. As an analogy, consider a piece of dust landing on a vibrating tuning fork, and the pitch of the fork changing slightly as a result. Achieving low photon loss in the resonator means that even tiny shifts in frequency can be detected. Such shifts are directly proportional to the strength of the perturbation or, in other words, to the number of particles that are in the vicinity of the resonator.

The sensing mechanism demonstrated by Hodaei et al. and Chen et al. was first proposed4 in 2014 (Fig. 1). The idea is to couple two or more modes of light in such a way that the modes and their corresponding frequencies coalesce. This occurs at points in the parameter space called exceptional points, which have been explored extensively in the context of photonic systems that exhibit a property called parity–time symmetry5,6.

Figure 1: Improving the sensitivity of optical-resonator sensors.
Figure 1

Structures called optical resonators trap light at certain frequencies. When the environment of such a resonator is perturbed, these frequencies shift, which allows optical resonators to be used as sensors. a, Hodaei et al.2 report a sensor that consists of three ring-shaped resonators that are coupled (red arrows). The authors use gold heating elements both to precisely tune the sensor and to emulate perturbations. b, By contrast, Chen et al.3 use a single toroidal resonator, and couple light that travels in clockwise (blue arrow) and anticlockwise (yellow arrow) directions. The authors use two fibre tips to tune the sensor and another type of tip to introduce perturbations. c, In conventional sensors, the shift in frequency caused by a perturbation is directly proportional to the strength of the perturbation (grey line). Hodaei et al. and Chen et al. demonstrate that the frequency shift in their sensing devices scales with the cube root (red line) or square root (blue line) of the perturbation strength, respectively. This leads to a dramatic improvement in the scaling of sensitivity of such sensors in comparison to conventional devices.

Exceptional points arise only in non-Hermitian systems, meaning that they do not exist in systems that conserve power. Therefore, whereas photon loss is a nuisance to be eliminated in conventional sensors, it is a key feature of sensors that contain exceptional points. Importantly, the frequency shift in the optical spectrum of an exceptional-point sensor is not directly proportional to the strength of the perturbation. Instead, for example, it scales with the square root or cube root of the perturbation strength — known as second-order and third-order exceptional points, respectively. This translates into a dramatic improvement in the sensing of tiny changes in the optical path when compared with conventional devices.

Hodaei and colleagues' sensor consists of three coupled, ring-shaped resonators based on the semiconductor indium gallium arsenide phosphide (Fig. 1a). The authors selectively injected light into the resonators through a process called optical pumping, such that one of the resonators lost photons, another gained photons and the third neither lost nor gained photons. A simple calculation shows that this set-up leads to the creation of a third-order exceptional point. However, making sure that the system lies at the exceptional point is a challenge for the same reason that the system makes a good sensor — it is extremely sensitive to small perturbations. To overcome this problem, Hodaei et al. placed a gold heating element under each of the resonators, which allowed the system to be precisely tuned. The authors then used the heaters to emulate perturbations to the system and demonstrated that the frequency shift in the optical spectrum scales with the cube root of the strength of the perturbation.

By contrast, Chen and colleagues used a single toroidal resonator made from silicon dioxide (Fig. 1b). Rather than coupling resonators, the authors coupled the modes of a given frequency of light that moved in clockwise and anticlockwise directions around a single resonator. They then used two nanometre-scale fibre tips to carefully tune the coupled system to an exceptional point. On the introduction of a perturbation in the form of another tip, the authors observed the square-root behaviour that indicates the presence of a second-order exceptional point.

The two research groups have married an abstract concept in physics (exceptional points) with concrete sensing applications, which represents a substantial accomplishment in the field of photonics. Owing to the universality of wave physics, many concepts in photonics can also be found in other fields, especially condensed-matter physics. Examples include photonic crystals7, Anderson localization (the absence of wave diffraction in a disordered medium) and exotic states of matter called topological insulators. However, because particle gain and loss are ubiquitous in optics (but not in condensed-matter physics, for example), non-Hermitian effects such as exceptional points provide optical scientists with a unique tool for probing fundamental physics and, in doing so, for discovering new applications.

There are still many issues to address. For instance, is there a way to avoid the trade-off between extreme sensitivity and the careful fine-tuning that is required for exceptional-point sensors? Can the sensing technique demonstrated by Hodaei et al. and Chen et al. be taken beyond proof-of-concept experiments to achieve higher sensitivities than those of the best low-loss optical-resonator sensors? And finally, what is the ultimate sensitivity limit of exceptional-point sensors?



  1. 1.

    & Nature Methods 5, 591–596 (2008).

  2. 2.

    et al. Nature 548, 187–191 (2017).

  3. 3.

    , , & Nature 548, 192–196 (2017).

  4. 4.

    Phys. Rev. Lett. 112, 203901 (2014).

  5. 5.

    & Phys. Rev. Lett. 80, 5243–5246 (1998).

  6. 6.

    et al. Nature Phys. 6, 192–195 (2010).

  7. 7.

    Sci. Am. 285, 46–55 (December 2001).

Download references

Author information


  1. Mikael C. Rechtsman is in the Department of Physics, The Pennsylvania State University, University Park, Pennsylvania 16802, USA.

    • Mikael C. Rechtsman


  1. Search for Mikael C. Rechtsman in:

Corresponding author

Correspondence to Mikael C. Rechtsman.

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing