Search tips
Search criteria 


Logo of scirepAboutEditorial BoardFor AuthorsScientific Reports
Sci Rep. 2016; 6: 28169.
Published online 2016 June 30. doi:  10.1038/srep28169
PMCID: PMC4928082

A self-sustaining atomic magnetometer with τ−1 averaging property

C. Xu,1,2 S. G. Wang,a,1,3 Y. Y. Feng,1,3 L. Zhao,1,4 and L. J. Wangb,1,2,3


Quantum measurement using coherent superposition of intrinsic atomic states has the advantage of being absolute measurement and can form metrological standards. One example is the absolute measurement of magnetic field by monitoring the Larmor precession of atomic spins whilst another being the Ramsey type atomic clock. Yet, in almost all coherent quantum measurement, the precision is limited by the coherence time beyond which, the uncertainty decreases only as τ−1/2. Here we show that by non-destructively measuring the phase of the Larmor precession and regenerating the coherence via optical pumping, the self-sustaining Larmor precession signal can persist indefinitely. Consequently, the precision of the magnetometer increases with time following a much faster τ−1 rule. A mean sensitivity of 240 An external file that holds a picture, illustration, etc.
Object name is srep28169-m1.jpg from 1 Hz to 10 Hz is realized, being close to the shot noise level. This method of coherence regeneration may also find important applications in improving the performance of atomic clocks.

Magnetometers based on atomic spin have achieved great progress in recent years and magnetometers with different characteristics find important applications in various fields1,2,3. High sensitivity below An external file that holds a picture, illustration, etc.
Object name is srep28169-m2.jpg has been demonstrated by Kominis et al.4 with a potassium SERF (spin exchange relaxation-free) magnetometer and by Budker et al.5 with NMOR (nonlinear magneto-optical rotation) magnetometers, making atomic magnetometers comparable to or even surpassing the state-of-the-art SQUIDs (superconducting quantum interference devices)6,7. In the field of ultrasensitive magnetic field detection, measurement of an external magnetic field can be realized by monitoring the Larmor precession of atomic spins8. However, the atomic spin will inevitably suffer from decoherence and decay to thermal equilibrium after a lifetime τatom, which limits the precision of the measurement. Hence, extending the coherence time is the key in improving the performance of all coherent quantum systems9,10. Here, we investigate and report a novel, self-sustaining atomic spin magnetometer based on coherent optical pumping. By synchronously switching on the pumping light at the Larmor precession frequency triggered by non-destructive measurement of its phase, the atomic spins can be regenerated coherently against the relaxation and thus maintain its precession indefinitely. The phase coherence time of the Larmor precession can thus be extended to a scale much longer than τatom. Consequently, the self-sustaining magnetometer utilizes the phase information, instead of merely the frequency of the Larmor precession to measure the magnetic field and shows the attractive feature that its measurement uncertainty averages down in time at a faster 1/τ rate.

The self-sustaining atomic magnetometer has several other advantages. It can self-oscillate from noise without requiring any initial preparation. Via the gyromagnetic ratio, it converts the measurement of magnetic field to that of frequency and time – which has the highest precision among all physical standards. Thus, this magnetometer is an absolute measurement device and can form a metrological standard. We note that spin precession driven by periodically modulated optical pumping was first studied by Bell and Bloom11 where persistent spin polarization was obtained. In the present work, we measure the precession spin signal’s phase via non-destructive measurement instead of by tracing the atoms’ absorption12,13,14, minimizing disruption to the precession itself. Finally, the continuous oscillating signal makes it much more convenient to apply further signal processing in many applications. For example, phase-locking of the precession signal to a reference local oscillator can produce a highly sensitive error signal to lock the magnetic field. The phase-locking stabilization of magnetic field can be very helpful when an ultra-stable magnetic field is needed8.

The magnetometer monitors the Larmor precession of the electron spin of an atom in the magnetic field. For a standard free precession process, the spin is first polarized along the propagation direction of a circularly polarized pump light, then the pump light is turned off and the spin starts precession around the external field  An external file that holds a picture, illustration, etc.
Object name is srep28169-m3.jpg . With the presence of decoherence, for example, caused by collision or by the atoms moving out of the detection region, the average value of the spin will relax to zero after a certain lifetime τatom. For a time interval τ > τatom the measurement can be repeated for a number of τ/τatom times, resulting in a sensitivity of1

An external file that holds a picture, illustration, etc.
Object name is srep28169-m4.jpg

where SNR is the signal-to-noise ratio and γ is the gyromagnetic ratio. If the system is limited by the spin projection noise only, then SNR equals An external file that holds a picture, illustration, etc.
Object name is srep28169-m5.jpg for an ensemble of N atoms. For a measurement time τ shorter than τatom, let τatom equal to τ. Then, the sensitivity will improve as 1/τ. For a measurement time τ much longer than τatom, the sensitivity improves as An external file that holds a picture, illustration, etc.
Object name is srep28169-m6.jpg due to the uncorrelated phase in each repeated measurement of time interval τatom. The polarization preparation by pumping in every measurement cycle destroys any phase coherence beyond time τatom, thus resulting in the An external file that holds a picture, illustration, etc.
Object name is srep28169-m7.jpg rule.

The spin self-sustaining method replenishes the atomic spin coherently. We illustrate the method using 85Rb atoms in the experiment as shown in Fig. 1(a). Setting the An external file that holds a picture, illustration, etc.
Object name is srep28169-m8.jpg -axis along the magnetic field, at t = 0 a circularly polarized light along the An external file that holds a picture, illustration, etc.
Object name is srep28169-m9.jpg -axis pumps all the atoms into the magnetic sublevel state |mFy = 3>. This pump light orients all atomic spins along the An external file that holds a picture, illustration, etc.
Object name is srep28169-m10.jpg-direction. Then the pump is turned off and the atomic spins precess and relax in dark. At time t = 2π/ωL the population will evolve back to the initial |mFy = 3> state. If the pumping light pulse is switched on right at this moment again, all atoms in |mFy = 3> will remain unaffected while atoms in all other states due to relaxation will be pumped back to |mFy = 3>. In this way, the spin is regenerated by the coherent pumping field and can maintain a very long lifetime.

Figure 1
Schematic diagram of the experimental apparatus.

The spin prepared via this method persists precession and behaves like a forced oscillator. The continuous signal can be described by An external file that holds a picture, illustration, etc.
Object name is srep28169-m11.jpg, where ΔA and Δϕ represent the amplitude and phase noise, respectively. ωL is the Larmor frequency. Owing to the spin self-sustaining method, the amplitude of the signal tends to be a constant value, reducing the amplitude noise. Hence, phase noise is the main factor that influences the measurement of An external file that holds a picture, illustration, etc.
Object name is srep28169-m12.jpg. The uncertainty Δt in the moment to turn on the pumping beam will mis-orient the spin with a phase error ωLΔt, introducing noise to Δϕ. Since each pumping moment is determined from the phase information of the Larmor precession signal measured before that moment, the phase of the Larmor precession in this method behaves like a “self-referencing clock”15. The uncertainty in each pumping moment will cause phase error to accumulate. In the situation when this cumulative phase error is small, the spin coherence is maintained, resulting in a τ1 reduction in magnetic field measurement uncertainty. When the accumulated phase error during a time interval τ0 is large enough to destroy the phase coherence, the measurement will then behave as An external file that holds a picture, illustration, etc.
Object name is srep28169-m13.jpg. The τ1 rule will change to a τ1/2 slope at time τ0, as analyzed in detail in ref. 15.


The configuration of the self-sustaining magnetometer is illustrated in Fig. 1(a,b). The signal of the probe light is proportional to the electron spin projection, <Sx>, along its propagation direction16 An external file that holds a picture, illustration, etc.
Object name is srep28169-m14.jpg. Due to the very large detuning set for the probe beam, it will only monitor the Larmor precession without directly influencing the atomic population in the states. This is a key feature of nondemolition measurement. We set the probe and pump beams in an orthogonal configuration such that when the spin precesses back along the pump light propagation direction, after one precession cycle, the signal of the probe just crosses the zero. This way, multiplicative amplitude error will be minimized. Coherent optical pumping of the spin is achieved by detecting the zero-crossing point with an analogue zero-crossing comparator and the rising edge of the output of the comparator is used as the trigger to turn on the pulsed optical pumping and repump light. This feedback loop synchronizes the pump with the Larmor precession. We’d like to point out that this synchronization technology can achieve high performance and synchronization of the rising edge by phase locking in a pure electrical system can be accurate to picosecond level.

Spin precession signals of the self-sustaining magnetometer are shown in Fig. 1(c,d). As shown from the signal observed in a “single-pump free precession” setup (d), the atomic spins have a lifetime of 30 ms, primarily limited by wall collisions. As shown in (c), the spin self-sustaining method keeps the atomic spins oscillating indefinitely.

For comparison, we test the performance of the magnetometer in both the “free-precession repeated” mode and the spin self-sustaining method, as shown in Fig. 2. In order to minimize the influence of magnetic field fluctuation, we monitor the coils current I with a 61/2 digital multimeter. The cross-correlation coefficient between the multimeter reading and the magnetometer measurement result is 0.94. Thus we can subtract the noise caused by the fluctuation in the coils’ current from the measured magnetic field B, through a linear relation An external file that holds a picture, illustration, etc.
Object name is srep28169-m15.jpg, where η is the conversion coefficient in unit of Tesla/Ampere.

Figure 2
Allan deviation of the “free-precession repeated” and spin self-sustaining method.


Timing errors may arise in the feedback loop. The possible noise comes from two parts: the noise in the Larmor precession signal itself and the one in the electronics. We need to reduce the intensity fluctuations in the probe light in the former part. In the latter part, one should use low noise photodetectors and filters. Great care must be taken to minimize the high frequency electronic noise in the high speed zero-crossing comparator since any noise in the comparator itself directly contributes to the timing error.

We analyze the sensitivity of our magnetometer by recording the continuous Larmor precession signal by a high speed 16-bit DAQ (data acquisition) card. The time sequence of zero-crossing points of the recorded signal is obtained with a linear fitting method near the zero point. The relative uncertainty in ΔT/T, with T being the period of Larmor precession of a single cycle, of each zero-crossing point due to data acquisition and processing is estimated to be below 5 × 10−5. Thus the main uncertainty in ΔT/T comes from the signal itself. Allan deviations17 shown in Fig. 2(a) display features clearly in accordance with the analysis. The self-sustaining method can retrieve magnetic field, B, from phase information of the Larmor precession. The phase noise Δϕ results in an uncertainty in B through An external file that holds a picture, illustration, etc.
Object name is srep28169-m16.jpg. When the cumulative phase error due to pumping is small, Δϕ is almost constant with time, leading to the 1/τ slope. The 1/τ rule can extend to about 300 ms, during which, the spin Larmor precession maintains phase coherence; τatom is only around 30 ms. It is followed by then turning into a An external file that holds a picture, illustration, etc.
Object name is srep28169-m17.jpg rule. The Allan deviation also shows the superiority of the spin self-sustaining method over the “free-precession repeated” mode on noise averaging property over time. The spin self-sustaining method can achieve a sensitivity of An external file that holds a picture, illustration, etc.
Object name is srep28169-m18.jpg, much reduced compared to the “free-precession repeated” method. The shot noise limit of spin projection according to Eq. (1) is estimated to be about 130 f T at An external file that holds a picture, illustration, etc.
Object name is srep28169-m19.jpg s for the experimental system.

The FFT (fast Fourier transform) spectrum, Fig. 2(b), supports the analysis above. The different slope in the An external file that holds a picture, illustration, etc.
Object name is srep28169-m20.jpg vs. An external file that holds a picture, illustration, etc.
Object name is srep28169-m21.jpg log plot in Fig. 2(a) corresponds to different noise types in the frequency domain with the form18 Sy(f) [proportional, variant] fα. Here, Sy(f ) represents the spectral density of the fractional frequency fluctuations y = Δf/f and α is the power-law exponent that depends on the noise type. For magnetometers, the Larmor precession frequency is proportional to B through the gyromagnetic ratio and thus Sy(f ) represents the spectral density of the relative magnetic field fluctuations. Since a An external file that holds a picture, illustration, etc.
Object name is srep28169-m22.jpg Allan deviation in time domain corresponds to white noise in the frequency domain with α = 0, the 1/τ Allan deviation will give a reduced noise in the low frequency range in frequency domain in Fig. 2(b) with19 α > 0.

Furthermore, we test the dynamic response of the magnetometer by applying sudden interruptions. This magnetometer can self-oscillate from the noise background without any initial pumping preparation and can regain self-oscillation after an interruption in the feedback loop in about 20 ms. Magnetometers of this type can quickly respond to sudden magnetic changes (60 nT within 1 μs, Fig. 3). Finally, we measure the Larmor frequency shift due to the circularly polarized pump light’s AC Stark effect20. The pulsed pump light leads to a much smaller effective intensity and reduced light shift. The Larmor frequency shift due to the pump light is as small as about 1.8 mHz/(mW/cm2), corresponding to about 380 f T/(mW/cm2) in the experiment.

Figure 3
(A) Black line: magnetometer starts from the noise background, plotted against the upper time axis. The self-sustaining feedback loop is turned on at high level (blue line). The red line is fitted to estimate the starting time (20 ms). (B) Black ...

In conclusion, we report a novel self-sustaining atomic spin magnetometer based on coherent optical pumping, demonstrate and investigate its features in details. As shown above, such a coherent pumping method can dramatically improve the performance of the magnetometer. The spin in this method behaves like a driven oscillator. The demonstrated 1/τ characteristics of noise level reduction via averaging in time using the self-sustaining method is very attractive and the method itself can be applied to other systems such as atomic clocks. The fast averaging, 1/τ rule enables magnetometers to reach the quantum shot noise floor of spin projection in a much shorter time. Also the magnetometer has a quick response to magnetic field variations and a reduced dependence on light shift. Since optical pumping can also be efficiently realized using spectral lamps, the magnetometer demonstrated here has a good prospect for miniaturization.

Additional Information

How to cite this article: Xu, C. et al. A self-sustaining atomic magnetometer with τ−1 averaging property. Sci. Rep. 6, 28169; doi: 10.1038/srep28169 (2016).


The authors thank Y. B. Gao for help in fabricating the coated cells and Y. H. Li and K. Wu for helpful discussions. The work is partially supported by National Natural Science Foundation of China under grants: 11204154, 11574016, 61473166 and China Postdoctoral Science Foundation funded project 2014M560958.


Author Contributions S.G.W. and L.J.W. conceived the experimental concept; All contributed to the design, and C.X., S.G.W. and Y.Y.F. conducted the experiment; C.X., S.G.W. and L.Z. analysed the data and C.X., S.G.W. and L.J.W. wrote the paper.


  • Budker D. & Romalis M. Optical magnetometry. Nature Physics 3, 227–234 (2007).
  • Kitching J., Knappe S. & Donley E. A. Atomic Sensors–A Review. IEEE Sens. J. 11, 1749–1758 (2011).
  • Shah V., Knappe S., Schwindt PDD. & Kitching J. Subpicotesla atomic magnetometry with a microfabricated vapour cell. Nature Photonics 1, 649–652 (2007).
  • Kominis I., Kornack T., Allred J. & Romalis M. A subfemtotesla multichannel atomic magnetometer. Nature 422, 596–599 (2003). [PubMed]
  • Budker D., Kimball D., Rochester S., Yashchuk V. & Zolotorev M. Sensitive magnetometry based on nonlinear magneto-optical rotation. Phys. Rev. A 62 (2000).
  • Koelle D. et al. . High-transition-temperature superconducting quantum interference devices. Reviews of Modern Physics 71, 631–686 (1999).
  • Zimmerman J. E., Thiene P. & Harding J. T. Design and operation of stable RF-biased superconducting point-contact quantum devices, and a note on properties of perfectly clean metal contacts. J. Appl. Phys. 41, 1572–1580 (1970).
  • Afach S. et al. . Highly stable atomic vector magnetometer based on free spin precession. Opt. Express 23, 22108–22115 (2015). [PubMed]
  • Bechtold A. et al. . Three-stage decoherence dynamics of an electron spin qubit in an optically active quantum dot. Nature Physics 11, 1005–1008 (2015).
  • Vijay R. et al. . Stabilizing Rabi oscillations in a superconducting qubit using quantum feedback. Nature 490, 77–80 (2012). [PubMed]
  • Bell W. & Bloom A. Optically driven spin precession. Phys. Rev. Lett. 6, 280–& (1961).
  • Alipieva E. et al. . Coherent population trapping for magnetic field measurements. Proceedings of SPIE. 5830, 170–175 (2004).
  • Brandt S., Nagel A., Wynands R. & Meschede D. Buffer-gas-induced linewidth reduction of coherent dark resonances to below 50 Hz. Phys. Rev. A 56, R1063–R1066 (1997).
  • Pollinger A. et al. . Control Loops for a Coupled Dark State Magnetometer. IEEE Sensors Conference, 779–784 (2010).
  • Wang L. J. On a New Class of Self-Referencing, 1/τ Atomic Clocks. Chinese Physics Letters 31 (2014).
  • Budker D. et al. . Resonant nonlinear magneto-optical effects in atoms. Reviews of Modern Physics 74, 1153–1201 (2002).
  • Allan D. W. The Statistics of Atomic Frequency Standards. Proc. IEEE 54, 221–230 (1966).
  • Leeson D. B. A simple Model of Feedback Oscillator Noise Spectrum. Proceedings of IEEE 54, 329–330 (1966).
  • Riley W. Handbook of Frequency Stability Analysis (NIST Special Publication 1065, Washington 2008).
  • Mathur B., Tang H. & Happer W. Light Shifts in Alkali Atoms. Phys. Rev. 171, 11–& (1968).
  • Budker D. & Kimball D. F. J. Optical Magnetometry (Cambridge University: Press, New York, 2013).

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group