Miniature cavity-enhanced diamond magnetometer

TL;DR

This paper introduces a compact, highly sensitive diamond-based magnetometer operating at room temperature, utilizing cavity enhancement and NV centers to achieve sub-picotesla sensitivity suitable for biomedical endoscopy.

Contribution

The work demonstrates a miniaturized cavity-enhanced diamond magnetometer with unprecedented sensitivity, advancing portable magnetic sensing technology for biomedical applications.

Findings

Magnetic-field sensitivity of 28 pT/√Hz achieved

Projected shot-noise-limited sensitivity of 22 pT/√Hz

Estimated quantum projection-noise-limited sensitivity of 0.43 pT/√Hz

Abstract

We present a highly sensitive miniaturized cavity-enhanced room-temperature magnetic-field sensor based on nitrogen-vacancy (NV) centers in diamond. The magnetic resonance signal is detected by probing absorption on the 1042\,nm spin-singlet transition. To improve the absorptive signal the diamond is placed in an optical resonator. The device has a magnetic-field sensitivity of 28 pT/$\sqrt{\rm{Hz}}$, a projected photon shot-noise-limited sensitivity of 22 pT/$\sqrt{\rm{Hz}}$ and an estimated quantum projection-noise-limited sensitivity of 0.43 pT/$\sqrt{\rm{Hz}}$ with the sensing volume of $\sim$ 390 $\mu$m $\times$ 4500 $\mu$m$^{2}$. The presented miniaturized device is the basis for an endoscopic magnetic field sensor for biomedical applications.

Full text

Introduction

Biomagnetic signatures are an important diagnostic tool to understand the underlying biological processes. Time-resolved biomagnetic signals are measured with Hall probes(Manandhar et al., 2009), Giant magnetoresistance sensors(Barbieri et al., 2016), alkali-vapor magnetometersJensen et al. (2016), superconducting quantum interference devices (SQUIDs)Ulusar et al. (2011) and single negatively-charged nitrogen-vacancy (NV) centers or ensembles thereofJ.F.Barry et al. (2016). Typical devices probe magnetic fields outside the body, i.e., far from their origin. However, signal strength and spatial resolution can both be improved by utilizing endoscopic sensors.

NV centers in diamond have already been used as nanoscale-resolution sensors Balasubramanian et al. (2008); Maze et al. (2008); Rittweger et al. (2009) with high sensitivity J.F.Barry et al. (2016); Wolf et al. (2015). Prominent examples of sensing with NV centers include, single neuron-action potential detection J.F.Barry et al. (2016), single protein spectroscopy Lovchinsky et al. (2016), as well as in vivo thermometry Kucsko et al. (2013). Due to their ability to operate in a wide temperature range as well as their small size, NV magnetometers are amenable for in-vivo and/or endoscopic applications.

The majority of NV sensors use a photoluminescence (PL) detection which suffers from low photon-detection efficiency. Approaches to counter this problem include, for example, the use of solid immersion lensesHadden et al. (2010); Siyushev et al. (2010); LeSage et al. (2012), or employ infrared (IR) absorptionDumeige et al. (2013); Acosta et al. (2010a); Jensen et al. (2014). Compared to NV sensors based on PL detection, those based on absorption feature collection efficiency approaching unityAcosta et al. (2010a).

Due to the small cross-section of the IR transition Acosta et al. (2010a); Dumeige et al. (2013), to achieve similar or higher sensitivities compared to PL-detection techniques, we use an optical cavity to enhance the optical pathlength in the diamond, and thus the IR absorption signal Dumeige et al. (2013); Jensen et al. (2014). With the cavity enhancement we can achieve sensitivities closer to the fundamental projection-noise limit, even at room temperatureDumeige et al. (2013); Jensen et al. (2014). Here we demonstrate a sensitive compact cavity-based IR absorption device operating near the photon shot-noise limit opening realistic prospects for a practical endoscopic magnetometer.

Experiment

The ground and excited electronic spin-triplet states of NV are 3A2 and 3E, respectively [Fig. 1 (a)], with the transition between them having a zero-phonon line at 637 nm. The lower and upper electronic singlet states are 1E and 1A1, respectively, with the transition between them having a zero-phonon line at 1042 nm (IR). While optical transition rates are spin-independent, the probability of nonradiative intersystem crossing from 3E to the singlets is several times higherWrachtrup and Finkler (2016) for $m_{s}=\pm 1$ than that for $m_{s}=0$. As a consequence, under continuous illumination with green pump light (532 nm), NV centers are prepared in the 3A2 $m_{s}=0$ ground state sublevel and in the metastable 1E singlet state. For metrology applications, the spins in the the 3A2 ground state can be coherently manipulated by microwave fields. In this work, the population of the ground state is inferred by monitoring IR light absorption on the singlet transition.

To increase the absorption of IR light we construct a cavity as shown in Fig. 1 (b). A spherical mirror with a curvature radius of 10 mm, reflectivity of 99.2(8)% (see supplemental materialSee Supplemental Material at [URL will be inserted by publisher] ) and diameter of 12.5 mm serves as the output coupler. A piezoelectric transducer is used to adjust the length of the cavity within a range of a few $\mu$m. It is glued with an epoxy resin (Torr Seal) between the spherical mirror and the ceramic holder for the diamond. The diamond plate serves as the input plane mirror of the cavity and is glued to the holder. The holder doubles as a heat sink. The (111)-cut diamond plate is dielectrically coated with high reflectivity $\sim$ 98.5% for IR light as well as anti-reflective for green light on the outside of the cavity. The diamond surface inside the cavity is supplied with an anti-reflective coating for both green light and IR light. The total optical length of the cavity is 5.00(3) mm, and the finesse is $\mathscr{F}$ = 160(4) (see supplemental materialSee Supplemental Material at [URL will be inserted by publisher] ). The cavity mode has a waist on the diamond with a radius of $38\,\mu$m; the mode radius is 54$\,\mu$m on the concave mirror surface. With this design, it is possible to bring the diamond’s outer surface in close proximity to a magnetic sample under study in compact geometry.

The setup for magnetometric measurements is shown in Fig. 1 (c). Green light is provided by a diode-pumped solid-state laser (Coherent, Verdi V10) and IR light is provided by an external-cavity diode laser (Toptica, DL-Pro). The green laser power is stabilized using an acousto-optical modulator (AOM, ISOMET-1260C with an ISOMET 630C-350 driver) controlled through a proportional-integral-derivative controller (PID, SIM960). The IR beam profile is matched to the lowest-order longitudinal cavity mode (TEM00), while the green beam is overlapped with the IR beam in the center of the cavity; it is not necessary to exactly mode-match the green beam profile. The frequency of the IR laser is locked to the cavity mode using a modulation technique with feedback to two PID controllers (SIM960). Fast feedback is realized by changing the laser current, while the cavity piezo actuator is used for slow feedback.

The microwaves (MW) to manipulate the NV spins are generated by a MW generator (SRS SG394). They are amplified with a 16 W amplifier (ZHL-16W-43+), passed through a circulator (CS-3.000, not shown in Fig.1) and high-pass filtered (Mini Circuits VHP-9R5), before they are applied to the NV centers using a mm-sized wire loop. The other side of the wire is directly connected to ground. A bias magnetic field is applied with a permanent ring magnet mounted on a precision positioning stage.

Results and discussion

The cavity transmission signal $I_{\textrm{tr}}$ for IR light is shown in Fig. 2 as a function of green light power $P$ in front of the cavity. The steady-state population of the singlet state increases with increasing green power, resulting in higher IR absorption. The IR absorption is enhanced by the cavity by 2$\mathscr{F}/\pi$, yielding significantly reduced IR transmission for higher pump powers. Higher absorption also results in an increase of the cavity-mode linewidth (Fig. 2, inset). The data in Fig. 2 are fitted with a saturation curve $I_{\textrm{tr}}\,=\,1-\alpha P/(P+P_{\textrm{sat}})$ (Ref.Jensen et al. (2014)), with saturation power $P_{\textrm{sat}}$ and reduction in transmission at saturation $\alpha$. The fit results are $P_{\textrm{sat}}$ = 735(1) mW and $\alpha$ = 0.605(1). Magnetic-resonance measurements (Fig. 3) are performed by scanning the MW frequency around the NV zero-field splitting (2.87 GHz). When the MW field is resonant with the ground-state $m_{s}=0$ $\rightarrow$ $m_{s}=\pm 1$ transitions, population is transferred through the exited triplet state to the metastable singlet state, resulting in increased IR absorption, which produces the observed optically detected magnetic resonance (ODMR) signal.

With a bias magnetic field (in this case, about 3 mT) aligned along the [111] axis, four peaks are visible by scanning the MW frequency (Fig. 3). The outer features result from the NVs along the [111] axis and the inner features from the remaining NV orientations. The contrast $\mathscr{C}$ and the full width at half maximum $\Delta\nu$ of the outer peaks are $\sim$ 3.7 $\%$ and 5.6 MHz, respectively.

For the magnetometric measurements we focus on the highest frequency feature in Fig. 3. We modulate the MW frequency $f_{\textrm{MW}}$ around the central frequency $f_{\textrm{c}}$ of the feature with frequency $f_{\textrm{mod}}$ = 8.6 kHz and modulation amplitude $f_{\textrm{dev}}$ = 4.5 MHz: $f_{\textrm{MW}}=f_{\textrm{c}}+f_{\textrm{dev}}\textrm{cos}(2\pi f_{\textrm{mod}}t)$ and detect the first harmonic of the transmission signal with a lock-in amplifier (LIA).

Fig. 4 shows the resulting dispersive signal centered at the feature (red) along with the feature itself (blue). Around the zero-crossing of the dispersive feature, we observe a linear signal S${}_{\textrm{LI}}\sim\alpha(f_{\textrm{c}}-f_{\textrm{res}})$ as a function of $(f_{\textrm{c}}-f_{\textrm{res}})$ when $\left\lvert f_{\textrm{c}}-f_{\textrm{res}}\right\rvert<<\Delta\nu/2$. We extract the slope $\alpha$ $\propto$ $\mathscr{C}$/$\Delta\,\nu$ from the fit of Fig. 4 (black) and use it to convert the magnetometer’s voltage output into magnetic field.

Figure 5 shows the magnetic-field-noise spectrum. The spectrum was obtained by a Fourier transform of the LIA output with a reference frequency of 8.6 kHz. The peaks at 50 Hz and harmonics are attributed to magnetic field from the power line in the lab and are not visible on the magnetically insensitive spectrum, which we obtain in the absence of a MW field. The noise floor in the region of 60-90 Hz for the magnetically insensitive spectrum is calculated as 28 pT/$\sqrt{\rm{Hz}}$. This sensitivity is $\approx$ 100 times better than what has been demonstrated previously with magnetometers based on IR absorption Jensen et al. (2014); Acosta et al. (2010a). Main improvements are: a dramatic reduction in cavity size, increase in probe laser power and improvements to the laser-lock stability. We verify the sensitivity by applying test magnetic fields (see supplemental materialSee Supplemental Material at [URL will be inserted by publisher] ). The photon shot noise limit is estimated as 22 pT/$\sqrt{\rm{Hz}}$ for 4.2 mW of collected IR light. The electronic shot noise is 2 pT/$\sqrt{\rm{Hz}}$. For an estimated NV density in the metastable singlet state of 0.68(1) ppm (see supplemental materialSee Supplemental Material at [URL will be inserted by publisher] ) and the demonstrated ODMR linewidth $\Delta\nu=5.6$ MHz we calculate a spin-projection noise limit of 0.43 pT/$\sqrt{\rm{Hz}}$. The bandwidth of the magnetometer is set by the LIA filter settings. For the presented measurements a time constant of $300\,\mu$s results in a 530 Hz bandwidth. The filter steepness is selected as 24dB/octave.

We demonstrate a miniaturized cavity-enhanced room-temperature absorption-based magnetometer using NV centers in diamond. The small size of our magnetometer yields a robust device with improved magnetic field sensitivity and makes it an ideal candidate for endoscopic measurements. The closer proximity to biomagnetic signal sources, inherent to endoscopic measurements, provides enhanced signal strength and spatial resolution, which may be further improved by using a combination of a different diamond and a higher finesse cavity. Our sensor features a noise-floor of 28 pT/$\sqrt{\rm{Hz}}$ close to the shot-noise limit (see supplemental materialSee Supplemental Material at [URL will be inserted by publisher] ). The sensitivity may be improved in future iterations by increasing the IR light powerAcosta et al. (2010b), using a critically matched cavityDumeige et al. (2013), implementing AC sensing protocols that allow increased NV coherence times due to dynamic decoupling from the decoherence sourcesFarfurnik et al. (2015), and using a diamond sample with narrower linewidth.

Acknowledgements.

We acknowledge support by the DFG through the DIP program (FO 703/2-1). GC acknowledges support by the internal funding of JGU. NL acknowledges support from a Marie Curie International Incoming Fellowship within the 7th European Community Framework Programme. LB is supported by a Marie Curie Individual Fellowship within the second Horizon 2020 Work Programme. DB acknowledges support from the AFOSR/DARPA QuASAR program. We thank J.W. Blanchard for a fruitful discussion.