A single beam Cs-Ne SERF magnetometer with differential laser power noise suppression method
Yao Chen, Yintao Ma, Minzhi Yu, Ning Zhang, Libo Zhao, Xiangguang Han, Zhuangde Jiang
AA single beam Cs-Ne SERF magnetometer with differentiallaser power noise suppression method
Yao Chen , Yintao Ma , Minzhi Yu , Ning Zhang , Libo Zhao , Xiangguang Han ,Zhuangde Jiang Abstract
We describe a single beam compact Spin Exchange Relaxation Free(SERF)magnetometer whose configuration is compatible with the silicon-glass bondingmicro-machining method. A cylindrical vapor cell with 3mm diameter and 3mm inlength is utilized in the magnetometer. In order to reduce the wall relaxation whichcould not be neglected in micro-machined SERF magnetometer, 3Amagats(1Amagat=2.69 × /cm ) neon buffer gas is filled in the vapor cell and thisis the first demonstration of a Cs-Ne SERF magnetometer. We also did a simulation toshow that neon is a better buffer gas than nitrogen and helium which is typical utilizedin vapor cells. In order to reduce the laser amplitude noise and the large backgrounddetection offset which is reported to be the main noise source of a single beamabsorption SERF magnetometer, we developed a laser power differential method and afactor of 2 improvement of the power noise suppression has been demonstrated. Finally,we did an optimization of the magnetometer and sensitivity of 40 f T /Hz / @30Hz hasbeen achieved. Introduction
Due to its high sensitivity, SERF magnetometer finds a wide range of application infundamental physics study and applied instruments. The application includesmagnetoencephalography(MEG) [1, 3, 4] andmagnetocardiography(MCG) [11, 18, 22],rotation sensing [5, 9, 13], testing physics beyondthe standard model: anomalous spin forces detection [8, 19, 20], low field NMR detectionin a micro-fluid chip [10],etc.The combination of atomic magnetometer with micro-machining technology couldfabricate chip-scale atomic magnetometers [1, 15]. With the merits of small size and lowcosts, chip-scale atomic magnetometer may find a wide range of application in theindustry. In the MEG study, smaller size means more detector could be fixed aroundthe head and this could give more precise source location of magnetic field produced bya bunch of neurons. In electrophysiology, better source location could give a better1/9 a r X i v : . [ phy s i c s . i n s - d e t ] J a n esolution of abnormal discharge area image before epilepsy surgery. Better resolution ofmagnetic field could be very important for the application of atomic magnetometer toBrain-Machine Interface.There are basicly two methods to fabricate an atomic magnetometer. One is thetraditional machining technology in which the alkali vapor cell is made of glass and theglass vapor cell is handled through a torch. The other method is the micro-machiningmethod in which the vapor cell is fabricated through glass-silicon anode bonding. It isobvious that the bonding technology can fabricate smaller vapor cell and the cost islower. The glass-silicon bonding cell only owns two transparency glass window sides.The traditional pump-probe configuration is not suitable for the atomic magnetometer.Hence an alkali atom spin modulation method with absorption detection method wasdeveloped to configure a single beam SERF magnetometer [15].Even though the simplified configuration only need one beam, the sensitivity of thesingle beam SERF magnetometer is far less sensitive than that of a pump-probeconfigured magnetometer. The improvement the sensitivity of the single beamabsorption detection is important for chip-scale atomic magnetometer. It is reportedthat the main noise source of the single beam magnetometer comes from the laser powernoise which is originate from a very large background offset after the laser pass throughthe vapor cell [15]. Here we developed a laser power differential method to suppress thebackground offset and laser power noise simultaneously to improve the sensitivity of thesingle beam absorption SERF magnetometer. We achieved a factor of 2 improvement ofthe laser power noise suppression.Alkali metal Rb is utilized in most of the SERF atomic magnetometer. The workingtemperature of Rb magnetometer is around 423K at which Rb atom could react withtypical utilized boro-silicate glass in anode bonding. Moreover, Rb owns 2 isotopeswhich are approximately at the same level in abundance in nature. It is expensive toisolate the two isotopes. Compared with Rb atoms, the working temperature of a CsSERF magnetometer is around 393K. The heated power consumption is lower than thatof Rb. Cs atoms are more difficult to react with glass at the same temperature thanthat of Rb. At the Cs working temperature, surface coatings such asOctadecyltrichlorosilane(OTS) could withstand such high temperature. The wallrelaxation which is much more obvious in micro-machined vapor cells could besufficiently suppressed in the magnetometers with OTS surface coatings. At last, thereis only one isotope in the nature abundance Cs metal. The cost of alkali vapor cellcould be reduced if Cs atom spins are utilized. Hench, we will focus on studying Csbased single beam SERF magnetometer.In a micro-machined SERF magnetometer, the size of an alkali vapor cell is typicallysmaller than 3mm in the three dimension directions which is much smaller than theoptical table magnetometer in which the vapor cell is larger than 1cm. The neglectedwall induced relaxation of atom spins is obvious. In order to suppress this relaxation,Nitrogen gas is typical utilized for it owns a large molecular diameter. However, thespin destruction rate between alkali atom and N is large. Thus, in this article, we choseNe gas as the buffer gas because the spin destruction cross section is 30 times smallerthan that of the N gas. Meanwhile, the diffusion constant of Cs in Ne is only twice asthat of the nitrogen gas. Compared with helium gas which needs special glass to stopits leakage, the leakage rate of Ne would be smaller than that of helium gas for thediameter of Ne molecular is larger. Thus we will focus on studying Cs-Ne SERFmagnetometer and we filled 3 amagats Ne in the vapor cell to reduce the wall relaxationin the Cs-Ne magnetometer. 2/9 heory Since it is reported that the main noise source of a single beam absorption SERFmagnetometer is from the laser power fluctuation [15], the zero field magnetic resonancelines could be narrowed to increase the sensitivity of the magnetometer. The line-widthof the magnetometer is directly decided by the transverse relaxation of the atomic spins,thus we need to reduce the spin relaxation. In a SERF magnetometer, the electronspins are under spin exchange relaxation free regime. The main relaxation comes fromthe wall relaxation and the spin destruction relaxation. The basic operation of a SERFmagnetometer could be found in this reference [2]. In a large spherical vapor cell whosediameter is 2.5cm, the wall relaxation can be neglected if 1 Amagat of buffer gas such asN , He and Ne is filled. In this paper, we focus on micro-machined alkali vapor cell andthe cell dimension is smaller than 3mm × × is typical in a micro-machinedvapor cell for the diffusion constant of alkali metal in nitrogen is large [15]. However,large nitrogen gas number density could cause substantially spin destruction relaxation.We find that Ne is better than N and we have done a simulation to decide which is thebetter choise for micro-machined vapor cells.Suppose that the laser beam is at the center of the cell and the beam diameter is0.8mm. The dimension of the cylindrical alkali vapor cell which is the typicalmicro-machined cell’s size is φ mm × mm . The theory of wall relaxation in acylindrical cell could be found in this paper [6]. As the atomic spins moved to thesurface of the vapor cell, their spin directions are totally randomized. The diffusion wallrelaxation time T D is: 1 ÷ T D = q ( µ + ν ) D where we have set µa = 2 .
405 and a is the diameter of the cylindrical vapor cell. ν = π/L and L is the length of the vapor cell. The wall relaxation is enhanced by afactor of q because both of the electron and nuclear spin polarization is destroyed at thewall.Note that we could not find an experimental parameters for the Cs-Ne spindestruction cross section, instead, we use a theoretical parameters of Cs- Ne spindestruction cross section from this paper [21]. The diffusion coefficient of Cs in Neongas can be find in this paper [7]. According to the theory analysis of a SERF atomicmagnetometer, total relaxation can be divided by the gyro-magnetic ratio of theelectron and 2 π to get magnetic resonance line-widths in the unit of nT [5]. Fig. 1 showsthe relationship between the buffer gas density and the equivalent magnetic resonanceline-width for several alkali metal-buffer gas pairs. From the results we conclude thatN owns a large diffusion coefficient which could efficiently stop the spins from wall spinrelaxation. However, with the increasing of buffer gas density, the spin destructionrelaxation becomes dominant. Compared with N , Ne owns a smaller diffusioncoefficient. However, the spin destruction relaxation rate is much smaller than that ofN at high gas pressure. For Ne, we find that the pressure is higher, the line-width isbetter. We chose the buffer gas pressure to be around 3 amagats which could be easilyreached when fabricating an alkali vapor cell. Above this pressure, it is not efficient toreduce the line-width. For Ne gas, the magnetic resonance line-width is 12nT. However,for N , the best pressure is around 1 amagat and the magnetic resonance line-width is20nT. therefore we can conclude that Ne is a better candidate for smaller alkali vaporcells. Hence we fill 3 amagats Ne in the alkali vapor cell described in this paper. 3/9igure 1. The equivalence magnetic resonance line-width resulted from wall relaxationand spin destruction relaxation for Cs atoms at different buffer gas densities.The theory of a single beam SERF magnetometer could be found in several articles.The pump-probe configuration is utilized in a traditional SERF magnetometer. In asingle beam SERF magnetometer, the absorption of the pump laser is detected tomeasure the external magnetic field. The absorbed laser power is proportional to theelectron spin polarization in the laser propagating direction. Since the magnetometer isonly sensitive to the polarization in the transverse plane which is perpendicular to thepumping light direction. A magnetic field modulation technique is utilized to increasethe sensitivity of the magnetometer as well as to reduce low frequency laser power noise.The theory and optimization method of the modulation could be found in severalreferences [16, 17]. Experimental Setup
The basic configuration of the experiment is shown in Fig. 2. The cylindrical vapor cellwhose inner diameter is 3mm and length is 3mm is made from boro-silicate glass. Itcontains a small droplet of Cs metal and 3 amagats of Ne gas is filled into it to reducethe wall collisions relaxation and 50 Torr of nitrogen is filled to quench the randomphotons from decay of the excited state Cs atoms. The cell is heated to 393K through a3 Watts 1550 nm laser which could efficiently heat the cell as well as produces nomagnetic field. The cell is fixed in a PEEK oven for thermal insulation. To eliminateearth magnetic field and its fluctuation, 3 layers of µ -metal shields together with acylindrical Mn-Zn ferrite are utilized. The residual magnetic fields could be furthercanceled actively by a set of homemade tri-axial high precision coils inside the magneticshields. Besides, the coil is also utilized to modulate the magnetic field to increase thesensitivity of the single beam absorption SERF magnetometer. In this system, only oneDFB laser which is tuned to the Cs D1 line (895nm) and directed along the z axis isused. The λ/ λ/ FB Laser Heating LaserFunction
Generator
NI DAQ λ /2 PBS ND Photodiode
Photodiode
Lock-in amplifier
Cell
CoilsMagnetic Shields λ /4 GT FiberPort
Figure 2.
Figure 2. The experimental setup of the Cs-Ne SERF magnetometer.PBS:Polarization Beam Splitter; GT:Glan-Thompson Polarizer; ND:Neutral DensityFilter; DFB:Distribution Feedback Brag.
TIAPhotodiodePhotodiode Lock-in amplifier
Figure 3.
Figure 3. The principal of the differential circruit for laser background andnoise suppression.TIA:Trans-impedence Amplifier.reference beam, a neutral density filter is utilized reduce the laser power before it isreceived by the other photodiode. The differential circruit showed in Fig. 3 is composedof the two photodiodes. The current could be substracted if the laser power of the twobeam is equal. The circruit could suppress the background offset and laser power noisesimultaneously to improve the sensitivity of the SERF magnetometer. The differentialsignal is amplified through a TIA and then it is demodulated by a lock-in amplifier suchthat the output of the magnetometer is proportional to the input magnetic field [16].The response of the magnetometer to the magnetic field is perpendicular to thedirection of the pump beam. The magnetic field signal finally was acquired through aNI(National Instrument) DAQ(data acquisition card).
Results
The magnetic resonance line-width had been simulated in the last section. In thissection, we first do a measurement of the line-width. This measurement not only givesinformation about the total relaxation of the magnetometer, but also can be utilized tooptimize the magnetometer to a better laser power working point. The temperature ofthe vapor cell is heated to 385K. Since 3 amagats of Ne are filled in the vapor cell. Thebroadened laser absorption line-width is 27.6GHz. These parameters will give an opticaldepth of 2.2. A modulation magnetic field with frequency of 990Hz is added to the x5/9
80 -60 -40 -20 0 20 40 60 80
Magnetic field (nT) -400-300-200-1000100200300400 R e s pon s e ( A r b . U n it ) Experimental dataFitted curve
Figure 4.
Figure 4. The response of the magnetometer to a static x transverse magneticfield. S e n s iti v it y ( A r b . un it ) DifferenalSingal Beam
Figure 5.
Figure 5. The noise spectrums of the laser power for the single beam andthe differential signal.direction. The modulation amplitude is optimized. At certain optical power, themodulation amplitude is changed and the response of the magnetometer to DC xmagnetic field is recorded. When the response is the largest, the modulation amplitudeis the optimized one.We found that the optimized amplitude is 200nT peak to peak. After themodulation amplitude optimization, we did a zero field magnetic field resonance lineshown in Fig. 4 [14]. We change the transverse x magnetic field and the response of themagnetometer is recorded. A Lorentz curve is fitted and we can get the line-width ofthe curve to be 30nT. This line-width equals to a total relaxation of 5275 sec − ,including the pumping rate, wall relaxation, Cs-Ne spin destruction relaxation andCs-Cs spin destruction relaxation. We changed the power of the laser then recordedseveral line-widths at different laser power. Then we extrapolate the line-width to zeropower point and we get a line-width of 16nT. Since the magnetometer responses largestif the electron polarization is 50%, we need to set the laser power of the magnetometerto a point when the line-width is 32nT. At the best laser power point, we did severalmagnetic field sensitivity measurements.Since the absorption of the pumping laser is detected in the single beam SERFmagnetometer, there is a background offset in the detected signal. The reference laser isutilized to do a differential measurement to suppress the background offset. Moreover,the laser power noise at the modulation frequency can also be suppressed in thisdifferential method. The diameter of the laser in our experiment is 0.8mm. We adjustthe reference laser power attenuation and the transmitted 22 µ A background currentcould be suppressed to be smaller than 500nA. In order to evaluate the differentialmethod, we measured the noise spectrums of the laser power with the differential 6/9 Frequency (Hz) Magnetometer SignalElectronic Noise
Figure 6.
Figure 6. The sensitivity of the magnetometer.method and without the differential method. The red dotted line in Fig. 5 shows thenoise spectrum of the laser power noise without the differential method. The blue solidline is the noise spectrum of the laser power with differential method. Note that thesignals are taken when the frequency of the laser is far from the absorption center andthere is nearly no interaction between the laser light and the atoms. As shown in Fig. 5,there is a factor of 2 improvement at the range of the magnetometer working frequencyrange.The laser power noise at 990Hz is suppressed by a factor of 2 in our experimentsetup.After optimization of the laser power and the modulation amplitude, we measuredthe sensitivity of our magnetometer. First we need to acquire the scale factor. We givea standard DC magnetic field in the x direction, then the output voltage change isacquired. In order to get the sensitivity, we adjust the DC input magnetic field to 0,then the voltage signal is recorded. We did a power density analyse of the noise signal.In order to give a more accurate result, we averaged the noise spectrum in the 1Hzrange. Fig. 6 shows the spectrum of the magnetic field noise. The resolution of thefrequency axis is 1Hz. We also shut off the laser then did a similar spectrum analyse forthe electronics and the photo-diode dark current noise. Note that the scale factor isfrequency depended because our magnetometer owns a finite bandwidth. The frequencyresponse of the SERF magnetometer is like an one order low pass filter [12]. The cutofffrequency [2] could be calculated through the spin relaxation rate. In our experiment,the zero magnetic resonance line-width is 30nT which corresponds to a total spinrelaxation of 5275 sec − . Together with the polarization of 0.5 for Cs atom spins, thecutoff frequency is calculated to be 516Hz. From Fig. 6 we can get that the sensitivityof our magnetometer is 40 f T /Hz / @30Hz. Conclusion
In conclusion, we have developed a Cs-Ne SERF magnetometer which is speciallydesigned for the future atomic magnetometer chip with micro-machined fabricatingmethod. We compared nitrogen, helium and neon buffer gas. The neon gas is betterthan nitrogen which is typical utilized in current chip-scale atomic magnetometer. Sincethe silicon-glass bonding vapor cell only owns two transparent windows, the single beamabsorption spin detection method is typical utilized. In order to reduced the largebackground offset and the laser power noise. A differential method is developed and afactor of 2 improvement could be realized. We finally get a magnetic field sensitivity of7/90 f T /Hz / @30Hz. Compared with Rb atoms, Cs atoms work at a lower temperature.The typical boro-scilicate glass which is utilized in the anode bonding process couldwithstand the chemical corrosion of Cs atoms. The power consumption of themagnetometer should also be smaller. Acknowledgments
This work is supported by Zhejiang Lab under grant number 2019MB0AB02,2019MB0AE01 and 113009-AA2003, China Postdoctoral Science Foundation undergrant number 2020M683462 and 2019M662121 and Jiangsu Province Youth Foundationunder grant number BK20200244.
References
1. O. Alem, R. Mhaskar, R. Jim´enez-Mart´ınez, D. Sheng, J. LeBlanc, L. Trahms,T. Sander, J. Kitching, and S. Knappe. Magnetic field imaging withmicrofabricated optically-pumped magnetometers.
Optics express ,25(7):7849–7858, 2017.2. J. C. Allred, R. N. Lyman, T. W. Kornack, and M. V. Romalis. High-sensitivityatomic magnetometer unaffected by spin-exchange relaxation.
Physical ReviewLetters , 89(13):130801, 2002.3. A. Borna, T. R. Carter, J. D. Goldberg, A. P. Colombo, Y.-Y. Jau, C. Berry,J. McKay, J. Stephen, M. Weisend, and P. D. Schwindt. A 20-channelmagnetoencephalography system based on optically pumped magnetometers.
Physics in Medicine & Biology , 62(23):8909, 2017.4. E. Boto, N. Holmes, J. Leggett, G. Roberts, V. Shah, S. S. Meyer, L. D. Mu˜noz,K. J. Mullinger, T. M. Tierney, and S. Bestmann. Movingmagnetoencephalography towards real-world applications with a wearable system.
Nature , 555(7698):657, 2018.5. Y. Chen, W. Quan, S. Zou, Y. Lu, L. Duan, Y. Li, H. Zhang, M. Ding, andJ. Fang. Spin exchange broadening of magnetic resonance lines in ahigh-sensitivity rotating k-rb-21ne co-magnetometer.
Scientific reports , 6:36547,2016.6. W. Franzen. Spin relaxation of optically aligned rubidium vapor.
Phys. Rev. ,115:850–856, Aug 1959.7. D. Giel, G. Hinz, D. Nettels, and A. Weis. Diffusion of cs atoms in ne buffer gasmeasured by optical magnetic resonance tomography.
Optics Express ,6(13):251–0, 2000.8. W. Ji, Y. Chen, C. Fu, M. Ding, J. Fang, Z. Xiao, K. Wei, and H. Yan. Newexperimental limits on exotic spin-spin-velocity-dependent interactions by usingsmco5 spin sources.
Physical review letters , 121(26):261803, 2018.9. L. Jiang, W. Quan, R. Li, W. Fan, F. Liu, J. Qin, S. Wan, and J. Fang. Aparametrically modulated dual-axis atomic spin gyroscope.
Applied PhysicsLetters , 112(5):054103, 2018. 8/90. R. Jim´enez-Mart´ınez, D. J. Kennedy, M. Rosenbluh, E. A. Donley, S. Knappe,S. J. Seltzer, H. L. Ring, V. S. Bajaj, and J. Kitching. Optical hyperpolarizationand nmr detection of 129xe on a microfluidic chip.
Nature communications ,5:3908, 2014.11. Y. J. Kim, I. Savukov, and S. Newman. Magnetocardiography with a 16-channelfiber-coupled single-cell rb optically pumped magnetometer.
Applied PhysicsLetters , 114(14):143702.1–143702.4, 2019.12. I. K. Kominis, T. W. Kornack, J. C. Allred, and M. V. Romalis. A subfemtoteslamultichannel atomic magnetometer.
Nature , 422(6932):596–599, 2003.13. T. W. Kornack, R. K. Ghosh, and M. V. Romalis. Nuclear spin gyroscope basedon an atomic comagnetometer.
Phys. Rev. Lett. , 95(23):230801, 2005.14. M. P. Ledbetter, I. M. Savukov, V. M. Acosta, D. Budker, and M. V. Romalis.Spin-exchange-relaxation-free magnetometry with cs vapor.
Phys. Rev. A ,77:033408, Mar 2008.15. V. Shah, S. Knappe, P. D. Schwindt, and J. Kitching. Subpicotesla atomicmagnetometry with a microfabricated vapour cell.
Nature Photonics , 1(11):649,2007.16. V. Shah and M. V. Romalis. Spin-exchange relaxation-free magnetometry usingelliptically polarized light.
Physical Review A , 80(1):013416, 2009.17. D. Sheng, A. R. Perry, S. P. Krzyzewski, S. Geller, J. Kitching, and S. Knappe. Amicrofabricated optically-pumped magnetic gradiometer.
Applied Physics Letters ,110(3):031106, 2017.18. I. A. Sulai, Z. J. Deland, M. D. Bulatowicz, C. P. Wahl, and T. G. Walker.Characterizing atomic magnetic gradiometers for fetal magnetocardiography.
Review of entific Instruments , 90(8):085003, 2019.19. W. Terrano, E. Adelberger, J. Lee, and B. Heckel. Short-range, spin-dependentinteractions of electrons: A probe for exotic pseudo-goldstone bosons.
Phys. Rev.Lett. , 115(20):201801, 2015.20. K. Tullney, F. Allmendinger, M. Burghoff, W. Heil, S. Karpuk, W. Kilian,S. Knappe-Gr¨uneberg, W. M¨uller, U. Schmidt, A. Schnabel, F. Seifert,Y. Sobolev, and L. Trahms. Constraints on spin-dependent short-rangeinteraction between nucleons.
Phys. Rev. Lett. , 111(10):100801, 2013.21. T. G. Walker. Estimates of spin-exchange parameters for alkali-metal ˘ noble-gaspairs. Physical Review A , 40(9):4959–4964, 1989.22. R. Wyllie, M. Kauer, G. S. Smetana, R. T. Wakai, and T. G. Walker.Magnetocardiography with a modular spin-exchange relaxation-free atomicmagnetometer array.