RESEARCH ARTICLE | JULY 12 2024

High dynamic-range and portable magnetometer using

ensemble nitrogen-vacancy centers in diamond 
Himanshu Kumar ; Shishir Dasika ; Maheshwar Mangat ; Siddharth Tallur ; Kasturi Saha 

Rev. Sci. Instrum. 95, 075002 (2024)

[Link]


View Export
Online Citation

Articles You May Be Interested In

Engineering an optical microphone for consumer electronics

J. Acoust. Soc. Am. (October 2020)

Enhanced magnetic field probe array for improved excluded flux calculations on the C-2U advanced beam-
driven field-reversed configuration plasma experiment
Rev. Sci. Instrum. (August 2016)

A high-density customizable microwave vacuum feedthrough for cryogenic applications

02 September 2025 [Link]

Rev. Sci. Instrum. (January 2020)
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

High dynamic-range and portable magnetometer

using ensemble nitrogen-vacancy centers
in diamond
Cite as: Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105
Submitted: 4 March 2024 • Accepted: 21 June 2024 •
Published Online: 12 July 2024

Himanshu Kumar,1,a) Shishir Dasika,1 Maheshwar Mangat,1 Siddharth Tallur,1,2

and Kasturi Saha1,2,3,b)

AFFILIATIONS
1
Department of Electrical Engineering, Indian Institute of Technology Bombay, Powai, Maharashtra 400076, India
2
Center of Excellence Semiconductor Technologies (SemiX), Indian Institute of Technology Bombay, Powai, Maharashtra
400076, India
3
Center of Excellence in Quantum Information, Computing Science and Technology, Indian Institute of Technology Bombay,
Powai, Maharashtra 400076, India

a)
himanshuk@[Link]

02 September 2025 [Link]

b)
Author to whom correspondence should be addressed: kasturis@[Link]. URL: [Link]
kasturi-saha/

ABSTRACT
Nitrogen vacancy (NV) centers in diamonds have been explored for a wide range of sensing applications in the last decade due to their
unique quantum properties. In this work, we report a compact and portable magnetometer with an ensemble of NV centers, which we call the
Quantum MagPI (Quantum Magnetometer with Proportional Integral control). Our fully integrated compact sensor assembly and control
× 10 × 7 cm3 box and a 30 × 25 × 5 cm3 rack-mountable box, respectively. We achieve a bandwidth normal-
electronics fit inside a 10 √
ized sensitivity of ∼10 nT/ Hz. Using closed-loop feedback for locking to the resonance frequency, we extend the linear dynamic range to
200 μT (20× improvement compared to the intrinsic dynamic range) without compromising the sensitivity. We report a detailed performance
analysis of the magnetometer through measurements of noise spectra, Allan deviation, and tracking of nT-level magnetic fields in real-time.
In addition, we demonstrate the utility of such a magnetometer by real-time tracking of the movement of an elevator car and door opening
events by measuring the projection of the magnetic field along one of the NV-axes under ambient temperature and humidity conditions.
Published under an exclusive license by AIP Publishing. [Link]

I. INTRODUCTION biological samples.7–9 However, more widespread use of NV− cen-

ter sensors outside the laboratory requires portable and complete
Quantum sensors based on negatively charged nitrogen integration of the sensor assembly and electronics. The NV− sen-
vacancy centers in diamond (NV− ) centers have been extensively sor assembly size depends on various factors, such as the choice of
used to sense magnetic fields relevant to various scientific and indus- the laser, the microwave power used for excitation and appropriate
trial applications, such as monitoring large currents in the bus bar microwave components, the choice and arrangement of magnets for
of an electric vehicle1,2 and sensing magnetic fields in high pressure providing bias magnetic field, the properties of the diamond sub-
anvil cells.3 Such applications are enabled by the viability of NV− strate, and finally the choice of fluorescence detection optics and
centers for a wide operating temperature range, from sub-1–1000 control and auxiliary electronic components. Typically, there is a
K4 and pressures up to 130 GPa.3 Similarly, the high sensitivity size and sensitivity trade-off in such sensors. For example,
√ the sen-
achievable with NV− centers (∼1 pT)5,6 has made it possible to sitivity of the NV− based sensor improves by a factor of P, where
detect weak magnetic fields emanating from very small currents in P is the excitation laser power. The maximum power that could be

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-1
Published under an exclusive license by AIP Publishing
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

used is limited by the saturation intensity of such sensors, which is II. MATERIALS AND METHODS
∼1 mW/μm2 .10,11 Hence, a laser with higher power can improve the The NV− center is a color defect in the diamond crystal, with
sensitivity of the system at the cost of increasing its size and weight. a nearest-neighbor pair of a vacancy and a nitrogen atom substitut-
Similarly, collection efficiency can be improved by using optics with ing a carbon atom in the diamond lattice. Negatively charged NV
a higher numerical aperture, which in-turn increases the size of the centers (NV− ) are unique because their ground and excited states
sensor. In the past, extensive work has been done to make NV− cen- are spin-triplet states (S = 1), and their quantum state can be ini-
ters as portable and compact as possible while still maintaining the tialized and manipulated via optical pumping and microwave (MW)
high sensitivity of the√sensor. Wang et al. obtained a magnetic field fields.21 The ground state of NV− is a spin−1 (S = 1) system, with
sensitivity of 21 nT/ Hz using a sensor head with dimensions of ms = 0 and ms = ±1 separated by a zero-field splitting of ∼2.87 GHz.
4 × 4 × 3 cm3 .12 The most sensitive portable NV− magnetometer The NV− spins can be excited using a 532 nm green laser and emit
has been reported by Stürner et al.,13 which utilized a sensor head red fluorescence (637 nm) while radiatively decaying to the ground
with dimensions of 3 × 3 × 2 cm3 and reported a sensitivity of state. These transitions are spin conserving. In addition, there exists
0.34 nT/ Hz. The smallest NV− center-based magnetometer has
√
an alternate non-radiative decay path to the ground state, transfer-
been reported by Kim et al.,14 wherein the microwave source, pho- ring the spin population from ms = ±1 to ms = 0 state, which enables
todetector, and optical filters have been integrated into a very small optical polarization of the NV− to ms = 0 state. A MW field enables
area of 200 × 200 μm2 using CMOS √ technology. However, the state transfer of the population from ms = 0 to ms = ±1 at reso-
reported sensitivity is only 32 μT/ Hz due to the extremely small nance, which can be mapped by measuring the photo-luminescence
size of the sensing elements. Another approach to miniaturize the (PL) intensity (also termed optically detected magnetic resonance,
sensor and impart further mobility is to place the diamond on a ODMR).21 With the application of a magnetic field, Zeeman splitting
15–17
fiber head.
√ The best reported sensitivity using this approach of the degenerate states ms = ±1 can be introduced, which leads to a
(0.31 nT/ Hz) has been reported by Patel et al.15 change in the resonance frequency of ms = 0 to +1 or −1 transition.
The utility of portable NV− center based magnetometers for In addition, the interaction of electron spin with nitrogen nuclear
applications such as geological sensing or oil and gas sensing, are spin leads to further splitting into two (15 N) or three (14 N) hyperfine
limited by the dynamic range of the sensor. The dynamic range levels. In this work, we have utilized a chemical vapor deposition
of an NV− center in an open-loop configuration is ∼0.1Γ/γe ,18 (CVD) grown (100) orientation diamond sample DNV-B1TM (Ele-
where Γ is the NV− linewidth of the optically detected magnetic ment Six) of size 3 × 3 × 0.5 mm3 , ∼1.1 ppm 13 C concentration,
substitutional nitrogen concentration [Ns0 ] ∼0.8 ppm, and [NV− ]

02 September 2025 [Link]

resonance spectrum and γe is the electron gyromagnetic ratio.
On the other hand, the sensitivity of the sensor is inversely pro- concentration ∼0.3 ppm.
portional to Γ. This implies that it is desirable to have lower A 525 nm, 1.2 W compact multimode fiber-coupled laser
linewidth for better sensitivity; however, this compromises the (Lasertack GmbH FL-525-1200) is used for optical excitation. The
dynamic range of the sensor. For example, a typical linewidth output beam from the fiber is collimated with a fixed focus colli-
of 1 MHz results in a dynamic range of only ∼3.5 μT. The res- mator, and the fiber collimator is mounted on a 30 mm cage XY
onance shift due to temperature shifts in NV− corresponds to translator (Thorlabs CXY1QA) with the help of a mounting adapter
2.6 μT K−1 . Therefore, a change in temperature of 1.3 K can shift (Thorlabs AD11F). The diamond was glued to an Ω-shaped loop
the NV− sensor outside its dynamic range in the open loop config- antenna of diameter 1.5 mm printed on a thin substrate (Rogers
uration, a problem that is exacerbated by the effects of any stray, RO4350BTM ). The output beam diameter incident on the diamond
external magnetic fields. Closed-loop control has been explored sample is ∼100 μm, and the number of NV centers probed in this
for improving the performance of magnetometry. Wang et al.19 volume is ∼1.72 × 1011 . Two 25 mm lenses were used to focus
reported an improvement in the bandwidth of tracking the magnetic the light emitted from the diamond, followed by a 600 nm long
fields with closed-loop frequency-locking utilizing a fast hopping pass filter (Chroma Technology Corp. AT600lp) to reject the green
frequency source and driving the three hyperfine resonances of 14 N light. Finally, the red fluorescence is collected on a photo-detector
nuclear spins simultaneously. Clevenson et al.20 reported one of the (Thorlabs DET36A2). The complete assembly is cage-mounted to
initial demonstrations of NV center based closed-loop frequency- form a compact sensor head. A hundred-turn current-carrying coil
locking magnetometry, utilizing two-channel frequency locking to (7 cm radius) is used to apply a homogeneous bias field. To further
∣0⟩ → ∣−1⟩ and ∣0⟩ → ∣+1⟩ transitions. While the authors claimed improve the footprint of the device, a 3 × 3 array of permanent mag-
a high dynamic range of ≈ 500 μT, details of sensor head size, nets (Neodymium 1 cm × 1 cm × 2 mm) is symmetrically glued on
portability of the setup, and experimental characterization were not both sides of the sensor head.
reported. The NV spin resonance is driven by MW generated using a
In this work, we report a portable, feedback stabilized NV− compact MW synthesizer [Windfreak SynthHD(v2)], amplified by
sensor (dubbed Quantum MagPI: Quantum Magnetometer with three low-noise amplifiers (Mini-Circuits ZX60-P33ULN+, gain 8.8
Proportional Integral control) with sensor head dimensions of dB each) cascaded to provide a gain of 26.8 dB, followed by a circu-
10 × 10 × 7 cm3 and all other electronics housed √ in a 30 × 25 lator (Pasternack PE8432) to prevent damage due to reflected power
× 5 cm3 rack-mountable box. The unit has 10 nT/ Hz sensitiv- from the antenna. The MW carrier signal is frequency-modulated
ity and a 200 μT dynamic range, can be powered using a standard (FM) and controlled with a Python interface. The FM signal fre-
wall-plug, operates outside the laboratory in ambient conditions, quency is given by f = fc + fdev sin(2πf m t), where fc denotes the MW
and is resilient to changes in temperature and other non-magnetic carrier frequency, fdev denotes the maximum frequency deviation,
disturbances. and fm is the modulation rate. The controller is implemented with

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-2
Published under an exclusive license by AIP Publishing
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

FIG. 1. Schematic of the experimental setup.

the Red Pitaya STEMlab 125-14 FPGA board as illustrated in Fig. 1, and τ is the integration time (see Sec. I of the supplementary material
and an open-source library (PyRPL) is utilized to perform lock-in for detailed calculation). Note that the applied bias field leads to a
homodyne demodulation, data acquisition, and closed loop tracking field component that is perpendicular to the NV axis. This off-axis
of NV resonances. The fluorescence signal from the photo-detector field impacts the sensitivity along the NV axis. We evaluated the
is amplified using a transimpedance amplifier (Edmund Optics Nos. effect of this perpendicular bias field by calculating the shift in reso-
59–178) and measured by the Red Pitaya board. A high-pass filter nance frequency between the perpendicular (B– ) and parallel (B∥ )
is used to remove the DC offset in the signal. The demodulated in- components of the magnetic field. The ratio of the two derivatives

02 September 2025 [Link]

phase and quadrature outputs are read using the Red Pitaya scope ∂ν
( ∂B / ∂ν ) is ∼0.014 (see Sec. II of the supplementary material for
– ∂B∥
modules, and data are transferred to the computer over an Ether-
detailed calculations), and therefore, we can use γe to calculate the
net interface. For closed-loop tracking of the resonance frequency,
sensitivity. The maximum slope is calculated by fitting the ODMR
the demodulation phase is adjusted such that the entire signal is
curve with the derivative of a Lorentzian function [Fig. 2(c)] and
present in one of the in-phase/quadrature components. This signal
finding the point where the slope is maximum. For the ODMR spec-
is provided as input to a proportional-integral (PI) controller, which −1
trum shown in Fig. 2(c), the maximum slope is 2.20 nV √ Hz , σ
generates an error signal to lock to the resonance frequency. An ana-
log adder is used to shift the DC level of the modulating signal by an = 15 ± 3 μV, τ = 10 ms, and the sensitivity is 10 ± 3.5 nT/ Hz. The
amount equal to the output of the PI controller. shot-noise limited sensitivity is given by10

1 Δν
ηshot−noise = PF √ , (2)
γe C R
III. RESULTS AND DISCUSSION
The ODMR spectrum is obtained by sweeping the MW fre- where PF is the numerical factor resulting from the line shape of
quency in the range of 2800–2950 MHz and observing the in-phase the resonance, Δν is the linewidth, C is the ODMR contrast, and
component [Figs. 2(a) and 2(b)]. The signal in the quadrature com- R is the
√ photon detection rate. For the Lorentzian line shape, PF
ponent is minimized by adjusting the demodulation phase. There = 4/3 3. For our ODMR spectrum, C = 0.15%, Δν = 1.0 MHz, R
14
are four possible orientations of NV centers along each of the C–C = 7.5 × 10√ photons/s, and the photon shot-noise limited sensitivity
bonds in the diamond lattice. The bias magnetic field is applied such is 0.7 nT/ Hz. The contrast is limited by the optical power utilized
that all four pairs of NV centers have different resonance frequen- for excitation and the MW power used to drive the resonance. At
cies. The four pairs of resonances correspond to ms = 0 to ms = ±1 a fixed laser power, the contrast increases with an increase in the
transitions and three hyperfine transitions due to 14 N nuclear spins microwave power. However, the MW power-induced broadening
for each of the four crystallographic axes of the NV center ensemble limits the sensitivity improvement at higher MW powers. Increas-
in the diamond lattice. The sensitivity of the magnetometer charac- ing the optical power to saturation requires higher microwave
terizes the minimum detectable magnetic field after an integration power to drive the electron spin resonance. The maximum MW
time of 1 s and is given by8 power in our setup is limited by the saturation power of the MW
√ amplifier used in the experiment (17 dBm at 1 dB compression
σ τ point). Applying higher power requires high saturation power MW
ηesr = dVlockin , (1)
γe df ∣max amplifiers, which adds to the size and power consumption of the
system.
where σ is the standard deviation of the ODMR spectrum, γe is the The performance of the magnetometer is further analyzed with
electron spin gyromagnetic ratio, dVdflockin ∣max is the maximum slope, noise power spectral density (PSD) measurements [Fig. 3(a)]. PSD

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-3
Published under an exclusive license by AIP Publishing
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

density of√ the magnetometer in the 10–100 Hz frequency band is

9.52 nT/ Hz.
The sensitivity of the magnetometer is also determined by
applying a square-wave magnetic field profile as a test signal, as
shown in Fig. 3(b). A 100-turn current-carrying coil (5 cm radius)
is placed in the vicinity of the magnetometer, and the coil field
is calibrated to measure the projection of the magnetic field along
the probed NV center axis. This is performed by sweeping the DC
current in the test coil and observing the shift in the resonant
frequency of the ODMR spectrum. The calibration constant of the
coil is determined by performing a linear fit to the frequency shift
vs DC current in the coil and extracting the slope of the linear fit
(see Sec. IV of the supplementary material). The extracted slope is
123.8 kHz/A, which translates to a calibration constant of 4.4 μT/A
when scaled by the gyromagnetic ratio. The coil is driven by a
modulated current, generating a magnetic field with a peak-to-peak
amplitude of ∼50 nT, and the MW frequency is set to the zero-
crossing point of the ODMR spectra. Hundred synchronous 8-s time
traces were acquired at a sampling rate of 1.9 ksamples/s and offset-
corrected and averaged in the time domain. The fluctuation in the
detected magnetic field for the 1 s window is shown in the inset of
Fig. 3(a). The standard deviation of √ the fluctuation is 11 nT, which
translates to a sensitivity of 12 nT/ Hz.
The linear dynamic range of the magnetometer for the open
loop and closed-loop cases is compared in Fig. 3(c). The MW fre-
quency is set to resonance frequency for the closed-loop case, and

02 September 2025 [Link]

the lock-in output is processed by the PI controller block imple-
mented on the FPGA. The integral and proportional gains for the
closed-loop case are −50 s−1 and −2.5, respectively. The PI output
signal generates an error signal to shift the frequency of the MW sig-
nal generator to track the resonance frequency. The dynamic range
for the closed-loop case is ∼200 μT, corresponding to a 20× improve-
ment compared to the open-loop case at ∼10 μT. This improvement
in dynamic range is achieved without compromising sensitivity.
The dynamic range is limited by the maximum frequency devi-
ation ( fdev ) possible with the MW signal generator, in this case,
5 MHz. The dynamic range can be improved further by using signal
generators with higher fdev .
FIG. 2. (a) Demodulated FM ODMR spectrum. The spectrum is obtained at 1 kHz The sensor’s performance was further analyzed with Allan devi-
modulation frequency, −3 dBm MW power, 400 kHz maximum frequency devia- ation measurements. The previous methods [Eq. (1) and Figs. 3(a)
tion, 10 ms lock-in time constant, and 700 Hz high pass filter corner frequency. (b) and 3(b)] to calculate the sensitivity assume that the dominant noise
Normalized integrated ODMR spectrum of the FM ODMR shown in panel (a) (see
Sec. I of the supplementary material for details). (c) Zoomed-in view of the ODMR process in the integration time is white noise. However, drift in the
spectrum corresponding to a resonance peak in the range 2.845–2.860 GHz. The signal due to variation in temperature and other environmental con-
extracted linewidth of each hyperfine level is ∼1 MHz. ditions is not accurately accounted for in these measurements, and
therefore, we study the Allan deviation of the signal. The overlap-
ping Allan deviation measurement22 was performed for 10 min of
recording of signals in three different cases: the open-loop mag-
netically sensitive case, the open-loop magnetically insensitive case,
plots are obtained by setting the MW frequency at resonance (mag- and the closed loop magnetically sensitive case. Figure 4 shows vari-
netically sensitive), away from resonance (magnetically insensitive), ation in the sensitivity [product of √ the Allan deviation σ(τ) and
and without laser excitation (electronic noise). The magnetic noise square-root of the integration time τ] for various values of sig-
spectral density is obtained by taking the fast Fourier transform nal integration time τ. The Allan deviation plot for the magnetically
(FFT) of time trace data acquired and normalized (see Sec. III insensitive measurement, which captures noise and drift due to the
of the supplementary material), scaled by the zero-crossing slope laser source, shows a flat profile up to 100 s integration time, indi-
(ZCS), and the gyromagnetic ratio of the ODMR curve. Fifty time cating that the dominant noise mechanism is white in frequency.22
traces (each of 1 s duration) were acquired for the three settings: In magnetically
√ sensitive measurement, the flat portion of the plot is
magnetically sensitive, magnetically insensitive, and laser excitation ≈10 nT/ Hz, which agrees with the sensitivity obtained from previ-
turned off to acquire the electronic noise floor. The noise spectral ous measurements. For larger τ, σ(τ) deviates due to the dependence

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-4
Published under an exclusive license by AIP Publishing
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

FIG. 3. (a) Noise spectral density plot. Fifty time-traces were acquired at 1.9 ksamples/s, 10 ms lock-in integration time, 1 kHz modulation frequency, and fdev = 400 kHz
for magnetically sensitive, magnetically insensitive, and laser OFF (electronic noise) operating cases. (b) Magnetic field tracking with an externally applied test signal with
a square-wave profile. The MW frequency is set equal to the resonance frequency. The time-traces are acquired at 1.9 ksamples/s, 25 ms lock-in integration time, 1 kHz
modulation frequency, and a frequency deviation of 400 kHz. Inset: histogram and Gaussian distribution fit of signal recording for a duration of 1 s. (c) Dynamic range
comparison of measurements performed in closed-loop and open-loop configurations.

02 September 2025 [Link]

FIG. 4. Sensitivity of the magnetometer measured for various values of inte-
gration time√ (τ). Sensitivity is obtained as the product of the Allan deviation
σ(τ) and τ. The confidence intervals are calculated using χ 2 statistics.22 The
“sensitive” trace represents an open-loop magnetically sensitive measurement,
and the “insensitive” trace represents a magnetically insensitive measurement.
“PI output” trace is obtained from Allan deviation measurement of the PI controller
output in closed-loop magnetically sensitive measurement, and “error signal” is the
error between the feedback signal and set-point (zero-crossing point of the ODMR
spectra, i.e., resonance frequency).

FIG. 5. (a) Photograph of a magnetometer placed 60 cm away from the door of an

of resonance frequency on temperature (≈74 kHz/degK).23 The
elevator on the second level of a four-story building, along with the control elec-
Allan deviation measurement was also performed for the closed- tronics box. Inset: schematic of the sensor head with the magnet array used to
loop feedback case, which overlaps with the magnetically sensitive apply the bias field. (b) Temporal response of the magnetometer when the eleva-
open-loop case. The error signal from the PI controller has − 12 slope, tor moves from ground level to the third level of the building with the magnetometer
showing that the signal has a white phase noise,22 indicating that placed at the second level. The data are acquired at a sampling rate of 1.9 ksam-
the drifts are transferred to the PI output to reduce the error in the ples/s and a 40 ms lock-in integration time. A Gaussian filter with σ = 50 is applied
to raw data.
lock-in output.

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-5
Published under an exclusive license by AIP Publishing
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

To demonstrate an application of this portable sensor, the mag- AUTHOR DECLARATIONS

netometer was utilized for real-time detection of magnetic fields
Conflict of Interest
generated by the movement of an elevator from the ground level
to the third level and door opening events. The magnetometer was The authors have no conflicts to disclose.
placed on the second level, 60 cm away from the elevator door of
a four-story building in the Department of Electrical Engineering Author Contributions
building at IIT Bombay (each level is 3.35 m above the other, and the
elevator door height is 2.28 m). A photograph of this set-up is shown Himanshu Kumar: Data curation (lead); Formal analysis (lead);
in Fig. 5(a), and the inset shows the schematic of the sensor head, Investigation (equal); Methodology (equal); Software (lead); Valida-
where we have replaced the bias magnetic field coil with a 3 × 3 array tion (equal); Visualization (equal); Writing – original draft (equal);
of permanent magnets (Neodymium 1 cm × 1 cm × 2 mm) symmet- Writing – review & editing (equal). Shishir Dasika: Data curation
rically attached on both sides of the sensor head to further improve (supporting); Formal analysis (supporting); Investigation (support-
the portability of the device. The device was operated in closed-loop ing); Methodology (supporting); Software (supporting); Validation
feedback control mode. Figure 5(b) shows the temporal response of (supporting); Writing – review & editing (supporting). Maheshwar
the magnetometer as the elevator moves to the first level from the Mangat: Data curation (supporting); Investigation (supporting);
ground level. The change in a magnetic field is measured in real time. Methodology (supporting); Software (supporting); Writing – review
When the elevator reaches the second level, the largest change in the & editing (supporting). Siddharth Tallur: Investigation (support-
magnetic field is observed because the elevator car is nearest to the ing); Supervision (supporting); Validation (supporting); Visualiza-
device. In addition, a change is observed when the door opens due to tion (supporting); Writing – review & editing (supporting). Kasturi
the magnetic field generated by the door opening mechanism. Upon Saha: Conceptualization (lead); Data curation (supporting); For-
reaching the third level, the response returned to the same level as mal analysis (supporting); Funding acquisition (lead); Investigation
the first level since the distance from the device was approximately (supporting); Methodology (lead); Project administration (lead);
the same. A similar response is observed when the elevator returns Resources (lead); Software (supporting); Supervision (lead); Vali-
to the first level from the third level. dation (supporting); Visualization (supporting); Writing – original
draft (equal); Writing – review & editing (equal).

IV. CONCLUSION

02 September 2025 [Link]

DATA AVAILABILITY
In conclusion,
√ we have realized a novel portable magnetome- The data that support the findings of this study are available
ter with ∼10 nT/ Hz sensitivity while simultaneously extending the
from the corresponding author upon reasonable request.
dynamic range by locking the frequency to the NV resonance. We
characterized the magnetometer for metrics such as magnetic noise
density, sensitivity, linear dynamic range, and stability. The sensitiv- REFERENCES
ity can be further improved by engineering the diamond, laser noise 1
Y. Hatano, J. Shin, J. Tanigawa, Y. Shigenobu, A. Nakazono, T. Sekiguchi, S.
cancellation, and the efficient collection of photons emitted from the Onoda, T. Ohshima, K. Arai, T. Iwasaki, and M. Hatano, “High-precision robust
diamond. Our work paves the way for applying the portable mag- monitoring of charge/discharge current over a wide dynamic range for electric
netometer to a wide variety of industrial applications in the near vehicle batteries using diamond quantum sensors,” Sci. Rep. 12, 13991 (2022).
2
future. K. Kubota, Y. Hatano, Y. Kainuma, J. Shin, D. Nishitani, C. Shinei, T. Taniguchi,
T. Teraji, S. Onoda, T. Ohshima, T. Iwasaki, and M. Hatano, “Wide temperature
operation of diamond quantum sensor for electric vehicle battery monitoring,”
Diamond Relat. Mater. 135, 109853 (2023).
3
SUPPLEMENTARY MATERIAL A. Hilberer, L. Toraille, C. Dailledouze, M.-P. Adam, L. Hanlon, G. Weck, M.
Schmidt, P. Loubeyre, and J.-F. Roch, “Enabling quantum sensing under extreme
See the supplementary material for sensitivity calculation, pressure: Nitrogen-vacancy magnetometry up to 130 GPa,” Phys. Rev. B 107,
calibration, normalization details, and an estimate of the power L220102 (2023).
4
consumption of various blocks in the system. G.-Q. Liu, X. Feng, N. Wang, Q. Li, and R.-B. Liu, “Coherent quantum control of
nitrogen-vacancy center spins near 1000 kelvin,” Nat. Commun. 10, 1344 (2019).
5
J. M. Schloss, J. F. Barry, M. J. Turner, and R. L. Walsworth, “Simultaneous
broadband vector magnetometry using solid-state spins,” Phys. Rev. Appl. 10,
ACKNOWLEDGMENTS 034044 (2018).
6
I. Fescenko, A. Jarmola, I. Savukov, P. Kehayias, J. Smits, J. Damron, N. Ristoff,
K.S. acknowledges the financial support from DST Quest N. Mosavian, and V. M. Acosta, “Diamond magnetometer enhanced by ferrite flux
Grant No. DST/ICPS/QuST/Theme-2/2019/Q-58, IIT Bombay Seed concentrators,” Phys. Rev. Res. 2, 023394 (2020).
Grant No. 17IRCCSG009, and SERB Power Research Grant No. 7
K. Arai, A. Kuwahata, D. Nishitani, I. Fujisaki, R. Matsuki, Y. Nishio, Z. Xin, X.
SPG/2023/000063. The authors acknowledge the support from staff Cao, Y. Hatano, S. Onoda, C. Shinei, M. Miyakawa, T. Taniguchi, M. Yamazaki,
and access to facilities at the Wadhwani Electronics Lab (WEL), T. Teraji, T. Ohshima, M. Hatano, M. Sekino, and T. Iwasaki, “Millimetre-scale
Department of Electrical Engineering, IIT Bombay, for building the magnetocardiography of living rats with thoracotomy,” Commun. Phys. 5, 200
(2022).
system reported in this work. H.K. acknowledges the funding from 8
J. F. Barry, M. J. Turner, J. M. Schloss, D. R. Glenn, Y. Song, M. D. Lukin, H.
the Prime Minister’s Research Fellowship. The authors acknowledge Park, and R. L. Walsworth, “Optical magnetic detection of single-neuron action
the insightful discussions with Dr. Madhur Parashar, IIT Bombay, potentials using quantum defects in diamond,” Proc. Natl. Acad. Sci. U. S. A. 113,
and Professor Saikat Ghosh, IIT Kanpur. 14133–14138 (2016).

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-6
Published under an exclusive license by AIP Publishing
 Review of ARTICLE [Link]/aip/rsi
Scientific Instruments

9 16
N. W. Hansen, J. L. Webb, L. Troise, C. Olsson, L. Tomasevic, O. Brinza, J. D. Zheng, Z. Ma, W. Guo, L. Niu, J. Wang, X. Chai, Y. Li, Y. Sugawara, C. Yu,
Achard, R. Staacke, M. Kieschnick, J. Meijer, A. Thielscher, H. R. Siebner, K. Berg- Y. Shi, X. Zhang, J. Tang, H. Guo, and J. Liu, “A hand-held magnetometer based
Sørensen, J.-F. Perrier, A. Huck, and U. L. Andersen, “Microscopic-scale magnetic on an ensemble of nitrogen-vacancy centers in diamond,” J. Phys. D: Appl. Phys.
recording of brain neuronal electrical activity using a diamond quantum sensor,” 53, 155004 (2020).
Sci. Rep. 13, 12407 (2023). 17
A. Kuwahata, T. Kitaizumi, K. Saichi, T. Sato, R. Igarashi, T. Ohshima, Y.
10
A. Dréau, M. Lesik, L. Rondin, P. Spinicelli, O. Arcizet, J.-F. Roch, and V. Masuyama, T. Iwasaki, M. Hatano, F. Jelezko, M. Kusakabe, T. Yatsui, and M.
Jacques, “Avoiding power broadening in optically detected magnetic resonance Sekino, “Magnetometer with nitrogen-vacancy center in a bulk diamond for
of single NV defects for enhanced dc magnetic field sensitivity,” Phys. Rev. B 84, detecting magnetic nanoparticles in biomedical applications,” Sci. Rep. 10, 2483
195204 (2011). (2020).
11 18
T. Wolf, P. Neumann, K. Nakamura, H. Sumiya, T. Ohshima, J. Isoya, and S. Dasika, M. Parashar, and K. Saha, “Mapping AC susceptibility with quantum
J. Wrachtrup, “Subpicotesla diamond magnetometry,” Phys. Rev. X 5, 041001 diamond microscope,” Rev. Sci. Instrum. 94, 053702 (2023).
19
(2015). C. Wang, Q. Liu, Y. Hu, F. Xie, K. Krishna, N. Wang, L. Wang, Y. Wang, K.
12
X. Wang, D. Zheng, X. Wang, X. Liu, Q. Wang, J. Zhao, H. Guo, L. Qin, J. C. Toussaint, Jr., J. Cheng, H. Chen, and Z. Wu, “Realization of high-dynamic-
Tang, Z. Ma, and J. Liu, “Portable diamond nv magnetometer head integrated range broadband magnetic-field sensing with ensemble nitrogen-vacancy centers
with 520 nm diode laser,” IEEE Sens. J. 22, 5580–5587 (2022). in diamond,” Rev. Sci. Instrum. 94, 015109 (2023).
13 20
F. M. Stürner, A. Brenneis, T. Buck, J. Kassel, R. Rölver, T. Fuchs, A. Savitsky, D. H. Clevenson, L. M. Pham, C. Teale, K. Johnson, D. Englund, and D. Braje,
Suter, J. Grimmel, S. Hengesbach, M. Förtsch, K. Nakamura, H. Sumiya, S. Onoda, “Robust high-dynamic-range vector magnetometry with nitrogen-vacancy centers
J. Isoya, and F. Jelezko, “Integrated and portable magnetometer based on nitrogen- in diamond,” Appl. Phys. Lett. 112, 252406 (2018).
21
vacancy ensembles in diamond,” Adv. Quantum Technol. 4, 2000111 (2021). L. Rondin, J.-P. Tetienne, T. Hingant, J.-F. Roch, P. Maletinsky, and V. Jacques,
14
D. Kim, M. I. Ibrahim, C. Foy, M. E. Trusheim, R. Han, and D. R. Englund, “Magnetometry with nitrogen-vacancy defects in diamond,” Rep. Prog. Phys. 77,
“A cmos-integrated quantum sensor based on nitrogen-vacancy centres,” Nat. 056503 (2014).
22
Electron. 2, 284–289 (2019). W. J. Riley and D. A. Howe, Handbook of Frequency Stability Analysis (National
15
R. Patel, L. Zhou, A. Frangeskou, G. Stimpson, B. Breeze, A. Nikitin, M. Dale, Institute of Standards and Technology, Gaithersburg, MD, 2008).
23
E. Nichols, W. Thornley, B. Green, M. Newton, A. Edmonds, M. Markham, R. Schirhagl, K. Chang, M. Loretz, and C. L. Degen, “Nitrogen-vacancy centers
D. Twitchen, and G. Morley, “Subnanotesla magnetometry with a fiber-coupled in diamond: Nanoscale sensors for physics and biology,” Annu. Rev. Phys. Chem.
diamond sensor,” Phys. Rev. Appl. 14, 044058 (2020). 65, 83–105 (2014).

02 September 2025 [Link]

Rev. Sci. Instrum. 95, 075002 (2024); doi: 10.1063/5.0205105 95, 075002-7
Published under an exclusive license by AIP Publishing
 1 Supplementary information: High dynamic-range and portable magnetometer using
2 ensemble nitrogen-vacancy centers in diamond

3 Himanshu Kumar,1, a) Shishir Dasika,1 Maheshwar Mangat,1 Siddharth Tallur,1 and

4 Kasturi Saha1, 2, 3, b)
5
1) Department of Electrical Engineering, Indian Institute of Technology Bombay, Powai,
6 Maharashtra, 400076, India
7
2) Center of Excellence in Quantum Information, Computing Science and Technology,
8 Indian Institute of Technology Bombay, Powai, Maharashtra, 400076,
9 India
10
3) Center of Excellence Semiconductor Technologies (SemiX),
11 Indian Institute of Technology Bombay, Powai, Maharashtra, 400076,
12 India

13 (Dated: 13 June 2024)

a) himanshuk@[Link]

b) kasturis@[Link]; [Link]

1
14 I. SENSITIVITY CALCULATION

15 In the sensitivity equation presented in equation (1) of the main text, the noise process is as-
16 sumed to be white. When the measurement bandwidth is limited by the sampling rate, the sensi-
17 tivity is given by1 :

√ p
18 η = σ T = σ / 2∆ f , (1)

19 where σ is the standard deviation of the 1 s long time trace, T is the sampling time, and ∆ f is the
20 Nyquist bandwidth. For lock-in measurement, the signal bandwidth is limited by the lock-in low
21 pass filter (LPF) bandwidth, which is smaller than the Nyquist limited bandwidth. The measured
22 σ is reduced in proportion to equivalent noise bandwidth. The sensitivity is given by:

√
ENBW /(2∆ f ) = σ ′ / 2EBNW ,
p
23 η =σ (2)

24 where σ ′ is the standard deviation of 1 sec long lock-in output time trace, and the equivalent
25 noise bandwidth (ENBW ) is the LPF bandwidth. The sensitivity from field tracking is obtained
26 by taking the standard deviation of the 1 s duration time trace. The sensitivity is obtained by
27 averaging N one second long time traces:

r
N
28 η =σ . (3)
2 × ENBW

29 To evaluate the shot-noise limited sensitivity, we need to obtain the contrast of the ODMR
30 spectrum, as is evident from equation (2) of the main text. We convert the FM ODMR spectrum
31 into a normalized ODMR spectrum by

i= j
∑i=1 xi
32 Snorm ( j) = ∆f, (4)
xbaseline 2 fdev

33 where Snorm ( j) is the normalized signal at jth MW frequency, xi is the lockin signal at ith MW
34 frequency point, xbaseline is the baseline signal measured by modulating the laser at the modulation
35 frequency, fdev is the maximum frequency deviation, and ∆ f is the MW frequency step size. The
36 contrast is obtained by calculating the difference between the maximum and minimum of Fig. 2(b)
37 in main text, and the shot-noise limited sensitivity is calculated using equation (2) in main text.

2
38 II. EFFECT OF OFF-AXIS FIELD

39 When the bias field is not aligned to a particular NV axis, the magnetometer is sensitive to the
40 component of the applied field perpendicular to the NV axis. Here, we evaluate the sensitivity
41 of our magnetometer to the parallel and perpendicular magnetic fields. The ground-state spin
42 Hamiltonian of NV center, neglecting the nuclear spin interactions, is given by2 :

H
43 = DSz2 + γe BNV Sz + γe (Bx Sx + By Sy ) + E(Sx2 − Sy2 ), (5)
h
44 where D is zero-field splitting, z is the NV quantization axis, γe is the electron gyromagnetic ratio,
45 BNV is the magnetic field parallel to the NV axis, Bx and By are the perpendicular magnetic field
46 components, E is the splitting due to the local strain obtained by zero-field ODMR. As shown
47 in Fig. (1a), the extracted value of E is 1.72 MHz. In our case, the farthest resonance peak
D
48 corresponds to the field of 1.85 mT, which is much smaller than γe = 100 mT, implies H⊥ << H∥ .
49 The magnetic field along each of the NV axis can be calculated by2 :

q
50 ν± = D ± (γe BNV )2 + E 2 , (6)

51 The unit vectors along the four NV axes in terms of lab-frame unit vectors are given by1
q q q q q q q q
52 n̂κ = ( 3 , 0, 3 ), n̂λ = (0, − 3 , − 3 ), n̂φ = (0, 3 , − 3 ), and n̂χ = (− 3 , 0, 13 ). The
2 1 2 1 2 1 2

53 resonance frequencies for each of the NV axes are extracted from ODMR spectra in Fig. 2(a) in
54 the main text. The total applied bias magnetic field |B0 | is obtained using the following equation:
4 4
55 ∆νκ2 − E 2 + ∆νλ2 − E 2 + ∆νφ2 − E 2 + ∆νχ2 − E 2 = γe2 (B2x + B2y + B2z ) = γe2 |B0 |2 , (7)
3 3
56 The value of |B0 | obtained from equation (7) is 1.96 mT, and BNV is 1.12 mT. The angle be-
57 tween the bias magnetic field and NV axis utilized for sensing is calculated by θB = cos−1 B|BNV0 | ,
58 and the perpendicular component of the magnetic field is calculated by B⊥ = |B0 |sin(θB ).
59 We sweep the perpendicular and parallel components of the magnetic field in the range 0.1 mT
60 (on the order of dynamic range) in positive and negative directions and solve the eigenvalues
61 of the Hamiltonian in the equation (5). The shift in the resonance frequency in response to the
62 parallel and perpendicular components of the applied field is shown in Fig. (1b). The deriva-
63 tive of frequency shift with the perpendicular magnetic field ( ∂∂Bν ) is −0.41 MHz mT−1 , and the
⊥

64 corresponding value for the the parallel field ( ∂∂Bν ) is 27.93 MHz mT−1 . The ratio of the two
∥

65 derivatives ( ∂∂Bν / ∂∂Bν ) is ∼ 0.014, and therefore the magnetometer is primarily sensitive to the
⊥ ∥

3
 (a) (b)

FIG. 1. Effect of off-axis magnetic field. (a) Zero field ODMR to calculate the local strain parameter (E),
(b) Resonance frequency shift due to the parallel (∆B∥ ) and perpendicular (∆B⊥ ) applied on the top of static
bias magnetic field.

66 parallel magnetic fields. Therefore, the error introduced by using γe for calculating the sensitivity
67 is significantly smaller considering the negligible effect of the perpendicular component.

68 III. NOISE SPECTRA

69 The noise spectra of the magnetometer shown in Fig. 3 of the main text are calculated by
70 normalizing the fast Fourier transform (FFT) of the signal:
s
2 N −i2π f n
x( f ) = x e fs , (8)
N fs ∑
71 n
1
72 where N is the total number of samples acquired in the time trace, fs is the sampling rate, and xn is
73 the nth sample. The averaged spectra corresponding to fifty traces are obtained as the square root
74 of the sum of squares of equation (8), and the calculations are verified by the Scipy Python library
75 periodogram function.

76 IV. TEST MAGNETIC FIELD COIL CALIBRATION

77 The current-carrying coil utilized to generate the test magnetic field is calibrated by sweeping
78 the DC current in the coil and performing ODMR at each sweep point. We placed the test magnetic

4
79 coil roughly 15 cm away, with the axis of the coil oriented at an angle of 45◦ with respect to the
80 beam. The ODMR spectra derivatives are fit using a derivative Lorentzian function:

 
1
81 F( f ) = α 1 − − f0 2
, (9)
1 + ( fΓ/2 )

82 where α is the amplitude of the ODMR curve, f0 resonance frequency, and Γ is the full-width-
83 half-maximum (FWHM) linewidth. The ODMR spectra obtained by varying the DC current in the
84 coil are shown in Fig. 2(a) and (b). Double-sided ODMR spectra were obtained for each current
85 to correct the drift in resonance frequency due to temperature changes. The resonance frequencies
86 are extracted and scaled by zero crossing (ZC) slope and electron spin gyromagnetic ratio, and
87 shown in Fig. 3. Linear fit is performed to obtain the calibration constant in order to map the
88 projection of the magnetic field along the probed NV center axis to current in the coil.

(a) (b)

FIG. 2. Test coil calibration. ODMR spectra obtained by sweeping the current in the coil. Derivative of
Lorentzian function was fit to obtain the resonance frequency. Spectra were obtained for both the transition
0 ↔ +1 and 0 ↔ −1 to cancel the temperature effects. (a) ODMR for 0 ↔ −1 transition, and (b) 0 ↔ +1
transition.

5
 FIG. 3. Calibration of magnetic field with current. The magnetic field increases as 4.4 µT/A.

89 V. ALLAN DEVIATION MEASUREMENT

90 The confidence intervals of Allan deviation measurement in Fig. 4 of the main text are obtained
91 by χ 2 statistics3 ,

ed f × s2
92 χ2 = , (10)
σ2

93 where s2 is the sample variance, ed f denotes the equivalent degrees of freedom. The term ed f
94 is determined by the number of points and type of noise3 . The Allan deviation measurements are
95 shown in Fig. 4(a). Fourier spectra of the Allan deviation data are normalized (equation (8)),
96 and shown in Fig. 4(b), (c) and (d) for magnetically sensitive, insensitive, and closed-loop cases,
√
97 respectively. As illustrated in Fig. 4(b) and 4(c), the flat portion of noise spectra is ∼ 10 nT/ Hz
98 confirming the sensitivity obtained from other methods.

6
 (a) (b)

(c) (d)

FIG. 4. Allan deviation measurement, (a) Allan deviation σ (τ) for three different cases: magnetically
sensitive - setting the microwave (MW) frequency at maximum slope point, insensitive - away from the
resonance, proportional-integral (PI) output and error signal for closed-loop case. (b-d) Noise spectral
density plot of the Allan deviation data for (b) magnetically sensitive case, (c) magnetically insensitive
case, and (d) for closed-loop case. The error signal and PI output show peaks around 200 Hz, indicating that
the system is underdamped.

7
 99 VI. ESTIMATION OF POWER CONSUMPTION

100 The calculated power consumption for various blocks in the apparatus are shown in Table I. The
101 power to the individual components is supplied by a master switch-mode power supply (SMPS)
102 of rating 24 V, 3 A. The total power consumption of the device under normal operating conditions
103 is roughly 41.3 W.

Sr. No. Component Voltage Current Power consumption

1 Laser driver 12 V 0.85 A (typ) 10.2 W

2 Peltier cooler 12 V 1.2 A (typ) 14.2 W

3 Microwave amplifiers 3V 1.72 A (typ) 4.16 W

4 Transimpedance amplifier ±15 V 0.11 A (typ) 0.33 W

5 Red Pitaya STEMlab 125-14 5V 2 A (max) 10 W

6 Cooling fan 12 V 0.2 A (max) 2.4 W

Total 41.29 W

TABLE I. Estimated power consumption of the sensor head and electronic control unit.

104 REFERENCES

105
1 J. M. Schloss, J. F. Barry, M. J. Turner, and R. L. Walsworth, “Simultaneous broadband vector
106 magnetometry using solid-state spins,” Phys. Rev. Appl. 10, 034044 (2018).
107
2 L. Rondin, J.-P. Tetienne, T. Hingant, J.-F. Roch, P. Maletinsky, and V. Jacques, “Magnetometry
108 with nitrogen-vacancy defects in diamond,” Reports on Progress in Physics 77, 056503 (2014).
109
3 W. J. Riley and D. A. Howe, “Handbook of frequency stability analysis,” (2008),
110 10.6028/[Link].1065.