ARTICLE

https://doi.org/10.1038/s41467-022-28169-z OPEN

Room-temperature optically detected magnetic

resonance of single defects in hexagonal boron
nitride
Hannah L. Stern1,4 ✉, Qiushi Gu1,4, John Jarman1,4, Simone Eizagirre Barker 1, Noah Mendelson2,
Dipankar Chugh3, Sam Schott1, Hoe H. Tan 3, Henning Sirringhaus 1, Igor Aharonovich 2 &
Mete Atatüre 1 ✉
1234567890():,;

Optically addressable solid-state spins are important platforms for quantum technologies,
such as repeaters and sensors. Spins in two-dimensional materials offer an advantage, as
the reduced dimensionality enables feasible on-chip integration into devices. Here, we
report room-temperature optically detected magnetic resonance (ODMR) from single
carbon-related defects in hexagonal boron nitride with up to 100 times stronger contrast than
the ensemble average. We identify two distinct bunching timescales in the second-order
intensity-correlation measurements for ODMR-active defects, but only one for those without
an ODMR response. We also observe either positive or negative ODMR signal for each
defect. Based on kinematic models, we relate this bipolarity to highly tuneable internal optical
rates. Finally, we resolve an ODMR fine structure in the form of an angle-dependent doublet
resonance, indicative of weak but finite zero-field splitting. Our results offer a promising
route towards realising a room-temperature spin-photon quantum interface in hexagonal
boron nitride.

1 Cavendish Laboratory, University of Cambridge, J.J. Thomson Ave., Cambridge CB3 0HE, UK. 2 ARC Centre of Excellence for Transformative Meta-Optical

Systems, Faculty of Science, University of Technology Sydney, Ultimo, NSW, Australia. 3 ARC Centre of Excellence for Transformative Meta-Optical Systems,
Research School of Physics and Engineering, The Australian National University, Canberra, ACT, Australia. 4These authors contributed equally: Hannah L.
Stern, Qiushi Gu, John Jarman. ✉email: [email protected]; [email protected]

NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z

D
efects in wide band-gap materials can host optically active that the single-defect ODMR contrast can reach beyond 30%,
confined spins that act as artificial atoms in convenient and approximately 100-fold stronger than the 0.4% contrast we
scalable platforms1,2. Colour centres in diamond3–5 and observe for the high-density ensemble measurements of the same
silicon carbide6,7 are prime examples of such systems, with long type of defect47. Strikingly, we also observe a bipolar ODMR
spin coherence times8 and high-fidelity spin control and read-out9. response across defects and this bipolar nature of the ODMR
Their coupling to nuclear spins further enables the realisation of contrast is explained by our kinetic model. Through second-order
optically accessible long-lived quantum memories1,10–14. Together intensity-correlation measurements per defect, we further show
with nanofabrication capabilities, these features make impurity that the presence of ODMR is correlated strongly with the pre-
spins leading candidates for light-based quantum information, sence of a second bunching timescale. Finally, below-saturation
sensing and communication technologies4,15–20. However, out- ODMR lineshape measurements and a spin model simulation
coupling light from defects in bulk crystals can be challenging and reveal that defects exhibit an angle-dependent doublet resonance,
most defects in bulk materials require low-temperature operation. consistent with a S > 1/2 system with modest zero-field splitting.
Layered van der Waals materials are an alternative Our results represent an important milestone for the develop-
platform21–25, where single-photon emitting defects are reported ment of room-temperature quantum optical platforms based on
to be among the brightest to date26 and the reduced dimen- individually accessible qubits in two-dimensional materials.
sionality may allow for a feasible route to designing scalable two-
dimensional quantum devices27,28. Hexagonal boron nitride
(hBN) is a two-dimensional van der Waals crystal that was Results
recently shown to host a plethora of defects that display sharp Material characterisation. To compare the behaviour of single
photoluminescence (PL) spectra at room temperature ranging hBN defects with the behaviour of previously reported defect
from 580 nm to 800 nm26,29,30, which can be tuned spectrally via ensembles, we measure a series of multilayer hBN films with
strain and electric field31–34. Multiple defect classes are emerging varying defect density, where the optical emission has been
in hBN: a structure involving a single negatively charged boron associated with carbon impurities47 (see Methods). The material
vacancy (VB−) displays broad emission at 800 nm and optically is grown via an MOVPE process that results in hBN layers with a
detected magnetic resonance (ODMR); however, this defect rough surface profile53 and clear wrinkles that can be seen in
has only been measured on the ensemble level35–38. There are confocal images (Supplementary Figs. 1, 2). The films show
also individually addressable defects around 700 nm, where the increasing levels of carbon-boron and carbon-nitrogen bonding
presence of spin has been inferred via their magneto-optical which in turn correlates with the defect density and brightness of
signature39, and recently via cryogenic ODMR measurements in the material under 532-nm illumination47.
crystalline hBN40. A family of narrow-band bright emitters with Figure 1 presents representative optical properties for the
distinctly sharper zero-phonon lines (ZPL) in the visible spectral defect ensemble in the high-density material (panel a) and for
range41–52 has recently received more attention; they can be defect A, a typical isolated defect in the low-density material
created controllably via chemical vapour deposition (CVD)44–47 (panels b and c). The insets of the panels a and b include
and plasma treatment methods48, display spectrally narrow bright integrated-PL intensity as a function of optical excitation power,
optical emission49, and have already been integrated into optical as well as integrated-PL confocal images showing the defect
cavities50–52. As such, they hold significant potential towards density for the two materials. In contrast to the broad PL
room-temperature devices for quantum-photonic applications; spectrum for the ensemble (Fig. 1a), the single-defect spectrum in
yet accessing their inherent spin at single-defect level is required Fig. 1b comprises well-resolved ZPL and multiple phonon
for their implementation as a room-temperature spin-photon sidebands (PSB) with an energy tuning of ~180 meV, consistent
interface. with previous reports47. Figure 1c is the non-background
In this article, we demonstrate that single defects in hBN host corrected second-order intensity-correlation measurement (g(2)
optically addressable spins at room temperature. We investigate (τ)) on the integrated-PL intensity for defect A. The antibunching
hBN with well-isolated single defects that have recently been behaviour shows g(2)(0) = 0.34(3) (Fig. 1c inset), indicating that
assigned to carbon impurities and show that they present strong defect A is an isolated single defect (Supplementary Fig. 11 for
optical signatures of single spins at room temperature. We find background-subtraction analysis).

Fig. 1 Optical properties of single and ensemble hBN defects. a Normalised PL spectrum of ensemble defects under 532-nm laser excitation. Dashed
vertical line represents the cut-off of a 550-nm long-pass filter. Inset: 5 × 5-µm image map of the integrated-PL intensity and example laser-power
dependence of PL counts at a fixed location. The colour bar scale is in units of million counts per second (Mcps). b Normalised PL spectrum of single defect
A under 532-nm laser excitation. Dashed vertical line signifies the 550-nm long-pass filter. Inset: 10 × 10-µm image map of the integrated-PL intensity with
a white circle around defect A, and the power dependence for defect A, showing optical saturation power (Psatoptical) = 70(2) µW and saturated PL
intensity (Isat) = 37,900(300) counts/s. Psatoptical reflects the power needed to achieve half the saturated PL counts (see Supplementary Fig. 3). The colour
bar scale is in units of thousand counts per second (kcps). c Second-order intensity-correlation measurement for defect A at 1.4Psatoptical excitation, solid
black curve is a theoretical fit. The fitted g(2)(0) = 0.34(3) and optical lifetime of 1.60(1) ns. Background analysis in Supplementary Fig. 11.

2 NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z ARTICLE

a b c

4 25 mT

ODMR Contrast (%)

2

4 Defect A
Ensemble
6 Defect B
0.5 0.6 0.7 0.8 0.9
Microwave Frequency (GHz)
d e
8 4

Frequency (GHz)
Defect A
3

ODMR
4
2
ODMR Contrast (%)

6
Contrast (%) 1
3 Defect A
4 0
0 100
Magnetic Field (mT)
2
2

1
0
Ensemble (x5)
0
0.01 0.1 1 10 100 0 1 2 3
P/Psat microwave ODMR Frequency (GHz)

Fig. 2 Room-Temperature ODMR Setup and Saturated ODMR Measurements. a An illustration of the measurement setup showing a permanent fixed
magnet positioned in-plane relative to the hBN layers with lithographically patterned microstrip. Inset schematic representing one cycle of the ODMR
protocol: a microwave pulse (orange) for the first half of the lock-in cycle, and signal and reference counts that are measured by the single-photon counting
detectors (APD) (blue). The excitation laser is present for the full lock-in cycle (green). b An optical image of a lithographically patterned 1-mm-long gold
microstrip used to apply microwave field to the hBN defects. c Room-temperature saturated ODMR of single defects (defect A, red circles; defect B, orange
circles) and of the ensemble (grey circles), all measured at 25-mT external in-plane magnetic field and at 10 times the microwave saturation power
(10Psatmicrowave). Psatmicrowave refers to the microwave power needed to achieve half the saturated ODMR contrast for the given defect or ensemble. The
solid curves are Lorentzian fits to the ODMR lineshapes. We determine a saturated linewidth of 34(3) MHz, 37(2) MHz and 34(2) MHz for the ensemble,
defect A and defect B, respectively. d ODMR contrast as a function of normalised microwave power (P/Psatmicrowave) for defect A (red) and the ensemble
(grey) with error bars showing one standard deviation. e ODMR spectra for defects A as a function of in-plane magnetic field (each colour represents a
different magnetic field strength), measured at 0.2 W, which is 10Psatmicrowave at 25 mT and 2Psatmicrowave at 89 mT. A constant microwave power was
used across the magnetic field range as higher powers cause microwave-induced heating. The black crosses, linked by a dashed line, mark the saturated
contrast at that magnetic field strength, i.e., 1.1% and 1.9% at 7 mT and 14 mT, respectively, saturating at ~4% at 25 mT and beyond (Supplementary
Fig. 27). The inset shows the ODMR resonance central frequency for each measurement shown in (e) (measurements plotted in the same colour) against
the magnetic field strength the measurement was performed with, fit to a linear function with a g-factor of 1.98(3).

Optically Detected Magnetic Resonance. Figure 2a illustrates the ~35-MHz linewidth. Strikingly, the single-defect ODMR signal
basic elements of our continuous-wave ODMR setup. We record has substantially higher contrast with respect to that of the
integrated-PL intensity under 532-nm laser excitation as a func- ensemble, up to 100-fold for some defects (Supplementary
tion of the applied microwave field. We modulate the amplitude Table 3). The comparable linewidth observed for the ODMR
of the microwave field with a square wave at 70 Hz to determine spectrum of the high-density ensemble, and the single defects
the difference between the PL intensity when the microwave field suggest that the mismatch might arise from a possibly low
is applied (signal) and when the field is not present (reference). fraction of spin-active defects, similar to previous reports39, as
The difference in PL is normalised by the reference PL intensity opposed to other effects such as spectral broadening of the
to obtain an ODMR contrast for each microwave frequency. This ODMR resonance. Indeed, out of more than 400 isolated defects
eliminates contributions from slow variations during each mea- we investigated for this work, 27 revealed measurable ODMR
surement. A permanent magnet mounted on a linear translation signal with fixed external magnetic field strength and orientation,
stage tunes the amplitude of the external magnetic field at the suggesting a yield in our experiments of ~5%. Further, ODMR
defect, which is applied in-plane relative to the hBN for the first signals of different signs are measured across different defects:
measurements presented. Figure 2b shows an optical image of one defects A and B in Fig. 2c are presented as examples of the
of our hBN devices, showing the lithographically patterned positive and negative ODMR contrast that we observe across the
microstrip on the hBN layer used to deliver the microwave field ODMR-active defects, with a roughly even yield of each polarity
locally. The microstrip is deposited on top of the grown hBN (Supplementary Table 3). A positive (negative) ODMR signal
multilayers, which uniformly span the image. indicates that microwave drive at spin resonance frequency leads
Figure 2c presents example ODMR spectra for the ensemble to an increased (decreased) PL intensity, which can further
(grey circles) and two single defects (red and orange circles) with contribute to the modest ODMR signal from the ensemble.
a 25-mT in-plane magnetic field, using a microwave field high Figure 2d presents the ODMR contrast of defect A as a function
enough to saturate the ODMR contrast. All three saturated of microwave power at 25-mT applied magnetic field, demon-
ODMR signals are at 700-MHz central frequency and show a strating the expected saturation behaviour. The ensemble ODMR

NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z

contrast shows equivalent saturation behaviour albeit at a (Supplementary Figs. 4–10), half of which show ODMR, and we
significantly lower ODMR signal. find a wide range in bunching timescales, consistent with the pre-
An ODMR frequency of 700 MHz at 25 mT is consistent with a vious reports26,39,43. However, interestingly we observe a strong
g-factor of ~2, typical for atomic spin defects in solids and Fig. 2e correlation between the presence of ODMR and the presence of two
presents the evolution of the ODMR spectra for defect A. The bunching timescales, independent of the laser power we use (Sup-
ODMR spectra in Fig. 2e are all acquired at a fixed input plementary Fig. 12). This is shown in panel c, where tri-exponential
microwave power (10Psatmicrowave at 25 mT), to compromise fits show that one of the two bunching timescales (denoted
between microwave-induced heating at high microwave field and τ b(additional) ) for the ODMR-inactive defects shows a significant
ODMR signal strength at low microwave field. The apparent σ 
variation of contrast, common to all defects, is due to the error τ b(additional) associated with the fit. This indicates that
frequency-dependent microwave transmission into the micro- b(additional)
while the photodynamics of ODMR-active defects is best described
strip. The black crosses and dashed line highlight the saturated
with two bunching timescales, non-ODMR defects display only one.
ODMR contrast for the corresponding spectra, which shows that
We find that the additional bunching timescale ðτ b(additional) Þ
the maximum ODMR response for defect A builds up to a steady
contrast of ~4% as a function of the magnetic field (Supplemen- ranges from 90 ns to 5.3 μs for the ODMR-active defects, which is
tary Fig. 19). The inset presents the magnetic-field-dependent shorter than the other bunching timescale (τ b ), which ranges from
shift of the central frequency for the ODMR signal for these 10 to 350 μs for all defects.
defects. A linear fit to the plot reveals a g-factor of 1.98(3) in line    
with the g-factor measured for other defects and with 2.03(3) ð2Þ jτt0 j=τ
ab jτt0 j=τ
b ð1Þ
g ðτÞ ¼ y0  ae þ be
measured for the ensemble (Supplementary Fig. 28).
   
jτt0 j=τ jτt0 j=τ
Bunching dynamics and the observation of ODMR. Second- g ð2Þ ðτÞ ¼ y0  ae ab þ be b
order intensity-correlation (g2(τ)) measurements were performed   ð2Þ
out to 1 ms time delays on the hBN defects to analyse the timescales jτt0 j=τ
þ ce b(additional)
associated with the optical transitions. In Fig. 3, g2(τ) measurements
are shown for defect B (panel a) and a second defect that did not
show ODMR (panel b). For both defects, the g2(τ) data is fit to bi- A simple three-level model with ground-state spin captures the
exponential and tri-exponential decay functions (Eqs. (1) and (2)) correlation between ODMR and two bunching timescales. In this
(tri-exponential not shown in (b)), which allows us to determine model the appearance of two bunching timescales arises from an
the antibunching (τ ab ) and bunching (τ b þ τ b(additional) ) time- imbalance in shelving and de-shelving rates between the spin
scales (panels d–f). We apply the same analysis across 18 defects sublevels of the optical manifold and the metastable state

Fig. 3 Bunching timescales of hBN ODMR-active defects. a Second-order intensity-correlation (g2(τ)) measurement of an ODMR-active defect, defect B,
measured at 1.5Psatoptical excitation (laser-power saturation in Supplementary Fig. 3), showing the dynamics out to 1-ms time delay ðt ¼ τ  t0 Þ, non-
background corrected. The data (circles) is fit to a bi- and tri-exponential fit (solid curves). b g2(τ) measurement of a defect that does not show ODMR, out
to 1-ms delay, measured at 0.2Psatoptical excitation, non-background corrected. The grey circles are the data and the solid line is a bi-exponential fit. For
background correction analysis see Supplementary Fig. 11. d–f The distribution of antibunching (τ ab ) and bunching (τ b(additional) and τ b ) timescales from
40 measurements of 18 defects. Data for defects that show ODMR is in red and defects that do not show ODMR in blue. Defects that don’t show ODMR
are not plotted in (e) because this data contains high error, as shown in (c). c A scatter plot (left plot) and histogram (right plot) of the error on the
 
σ
fractional error on the fit τ b(additional) of the additional bunching timescale, for ODMR and non-ODMR-active defects.
b(additional)
4 NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications
NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z ARTICLE

Fig. 4 Sub-linewidth structure and angle-dependence of below-saturation ODMR spectra. a Unsaturated (Psatmicrowave and 1.2Psatoptical) normalised
ODMR spectra for defect A with double-Lorentzian fits at a range of magnetic-field strengths. Coloured circles are the experimental results obtained at
7 mT, 14 mT, 25 mT and 89 mT, coloured solid curves are double-Lorentzian fits. For each panel, the shaded blue and red Lorentzian lineshapes show the
two components of the doublet. b The ODMR frequency of each component of the double-Lorentzian fit, as a function of magnetic field. c–h Below-
saturation normalised ODMR spectra for defects B, C, D, E, F and G obtained at 25 mT with a fixed in-plane magnet position. Shaded regions indicate the
two components of the doublet Lorentzian fits. (i) ODMR simulation for S = 1 system with D = 14 MHz and E = 4 MHz (j) ODMR simulation for S ¼ 32
system with D = 7 MHz and E = 2 MHz. In both (i) and (j) the magnetic field applied along the lab frame axis z and the defects symmetry axis (principal
axis of defect’s D tensor) z′. The signal intensity is normalised to 1 and represented by the blue shaded region.

(for detailed model see Supplementary Figs. 13–18). This rate plane magnetic-field strength for defect A and panel b displays
imbalance leads to ground-state spin polarisation, associated the central peak frequency for each Lorentzian of the doublet. The
generation of ODMR contrast, and the appearance of two splitting for defect A is ~30 MHz across the magnetic field range
bunching timescales. However, if all spin sublevels couple equally from 7 mT to 89 mT, and the average linewidth of the constituent
to the metastable state, the defects will show only one bunching single resonances is ~20 MHz. However, the measured doublet
timescale but no spin polarisation. Our model uses the same splitting varies across defects, between 19 and 50 MHz with a
three-level structure as other reports40,54, but it should be noted mean splitting of 34(8) MHz (Supplementary Table 3). This
that it also requires the addition of laser-power-dependent continuum of values suggests that the observation of singlet
shelving and de-shelving rates. resonances in some defects could be due to the presence of a
doublet with a splitting too small for us to resolve.
ODMR fine structure of hBN defects. To resolve sub-linewidth In principle, both crystal-field in the high-field regime and
features in the ODMR spectrum, we operate at low microwave hyperfine coupling can lead to a split doublet in the ODMR
driving conditions to avoid power broadening. As such, we spectrum. Electron paramagnetic resonance (EPR) measurements
operate in the near-optimal regime of signal strength with have shown that electronic spins in hBN couple to nitrogen,
microwave excitation power at Psatmicrowave, as inferred from carbon and boron nuclear spins35,55,56. However, the predicted
saturation measurements (Supplementary Fig. 19, 20). Figure 4 hyperfine constants and the corresponding splitting for boron55
presents the corresponding below-saturation ODMR spectra isotopes differ starkly from our results. Potential single-carbon
obtained with the external magnetic field applied in the plane of substitution defects, (CN and CB) are predicted to show
the hBN layers, for the defects labelled A to G. The coloured broadened resonances, rather than a distinct 30-MHz
circles are the data, and the solid curves are the Lorentzian fit splitting40,57. Coupling to one C13 nuclei could in principle
(Supplementary Figs. 21–25 for Gaussian and Voigt fit analyses). result in a doublet, however the abundance of C13 (~1%) does not
For most of the defects (~80%), we resolve a doublet structure, reconcile with the yield of ODMR-active defects we measure
while for others we cannot resolve a splitting. The corresponding (~5%)57. The hyperfine constant for nitrogen is in the correct
panels in Fig. 4 show the constituent individual lineshapes of the range35, but we do not expect a doublet spectrum from electron-
doublet resonances (shaded red and blue), obtained with a nitrogen coupling. All these make it difficult to assign the ODMR
double-Lorentzian fit. doublet to hyperfine coupling without considering a more
For the defects where we resolve doublets, the splitting is complicated atomistic structure.
independent of the magnetic-field strength. Figure 4a demon- An alternative origin for the ODMR doublet is zero-field
strates such independence of the doublet splitting from the in- splitting of a S > 12 state. To explore this possibility, we simulate

NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z

a b c
170°
z
135°
Norm. ODMR Contrast

Out-of-plane Angle
90°

MW field
45° sa
x pl m p
an l e
e
y
0°
100 50 0 50 100 100 50 0 50 100
Frequency Detuning (MHz) Frequency Detuning (MHz)
d e f

80°
z
Norm. ODMR Contrast

In-plane Angle 60°

MW field
sa
40° x p l mp
a n le
e y

100 50 0 50 100 100 50 0 50 100

Frequency Detuning (MHz) Frequency Detuning (MHz)

Fig. 5 Angular dependence of ODMR below saturation. a ODMR spectra (circles) for defect A at two magnetic field orientations, along the z-axis (red)
and in the x–y plane (orange), magnetic field vectors represented in (c). The solid curves are fits to the data using a S = 1 model with D = 25 MHz,
E = 5 MHz and the defect D tensor symmetry axis (z′) rotated in the plane of hBN. b The simulated ODMR contrast for defect A through the red plane
shown in panel (c). The shaded area represents signal intensity normalised to 1. c Schematic showing the magnetic field orientations (red and orange
vectors) for measurements in (a) and (b). The red vector points along z, and the orange points out of the xy plane. d ODMR spectra (circles) for defect E at
a series of magnetic field orientations through a plane, shown in panel (f). The solid curves are the fit the S = 1 model with D = 25 MHz, E = 5 MHz and z′
rotated in the hBN plane (e) The simulated ODMR contrast for defect E through the blue plane shown in panel (f). f Schematic showing the magnetic field
orientations (blue vectors) for measurements in (d) and (e). The blue vectors point at different angles in zy plane with a projection onto x. The sample
orientation is shown in both (c) and (f) but the cartoon image (in zy plane).

the expected ODMR spectra for both S = 1 and S = 32 systems, the dominating eigenvalue of the D tensor and denoted z′).
using low zero-field splitting parameters and an in-plane When z′ is aligned with the applied magnetic field, the splitting
magnetic field applied down the principal symmetry axis of the is determined by 2D, but if the defect is oriented with the
defect D tensor (panels i and j of Fig. 4). Neglecting hyperfine symmetry axis offset from the magnetic field, the splitting is
coupling, the spin state for a given defect can be described by a defined by D and E. In this case, with E = 5 MHz, this results in
spin Hamiltonian in the form, a splitting that cannot be resolved and instead appears as a
    single peak. Figure 5 shows results for two defects (panels a and
1 d) where the orientation of the external magnetic field is moved
H ¼ gμB B  S þ D S2z  SðS þ 1Þ þ E S2x  S2y ; ð3Þ
3 through a series of calibrated orientations relative to hBN plane
(vector of magnetic field shown in panels c and f). For these two
where μB is the Bohr magneton, S is the spin projection operator, defects, and all other defects measured (Supplementary
g is the g-factor, B is the external magnetic field, and D and E are Figs. 32–41), the ODMR lineshape shows a splitting that can
the zero-field splitting parameters. For S = 1 the two ODMR be tuned with angle and that is well described by the S = 1
transitions correspond to transitions between the ms = 0 and the model with D = 25 MHz and E = 5 MHz (simulation Fig. 5b, e).
ms ¼ ± 1 states and are separated by an energy of 2D. For a The largest splitting we measured, 50 MHz, corresponds to a
S = 32 system, the spin transitions that result in ODMR contrast defect with z′ along z (Supplementary Fig. 40). The only free
are ms = 32 to ms = 12 and ms = þ12 to ms = þ32, split by 4D in parameter in the model is the orientation of z′. We also note
energy58. As we can see in panels i and j, using slightly different D that for a small fraction of defects, where a doublet splitting was
and E values, both spin multiplicities can produce a doublet that not resolved, there was no noticeable dependence of the ODMR
does not change with magnetic field strength (for more details see lineshape on the magnetic field orientation (3 defects we
SI, section 5). measured). Experimental results for these defects are consistent
The appearance of both doublets and singlets in the ODMR with defects where z′ is tilted out of the 2D plane of the hBN
measurements can also be explained by a S > 12 model, if we sheets (Supplementary Figs. 37–39).
consider that not all defects will have the same orientation in
the hBN sample. For both S = 1 and S = 32 situations with low Discussion
zero-field splitting, the splitting of the ODMR doublet is For the ODMR-active defects, we observe some variation of the
dependent on the orientation of the external magnetic field contrast and lineshape, but the overall behaviour is remarkably
relative to the symmetry axis of the D tensor (determined by similar across the single defects studied with roughly equal

6 NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z ARTICLE

likelihood of finding positive or negative ODMR contrast sign. magnetic dipolar coupling parameter of electrons separated by
This bipolarity is unlike the defects in diamond that show a 1.3 nm, which corresponds to 10 bond lengths in hBN. This
consistent ODMR sign for a given optical defect whether probed small dipolar coupling, combined with poorly resolved hyper-
as a single or on the ensemble level. However, our analysis reveals fine coupling may be consistent with a carbon cluster defect and
a mechanism where all the ODMR behaviour we observe can be warrants further investigations to shed light on the atomistic
explained by a single type of optically active spin defect pre- structure.
senting highly tuneable photodynamics. Our kinetic analysis In conclusion, we report optically accessible spin defects in
shows that hBN defects display a wide range in bunching beha- hBN layers via ODMR measurements at room-temperature.
viour, but that the presence of two bunching timescales strongly We observe ODMR contrast for single well-isolated defects.
correlates with the presence of ODMR. In addition, we demon- The sub-unity yield of the ODMR-displaying defects, as well
strate that the ODMR contrast and sign can be determined by the as the polarity of the ODMR sign, are likely reasons for the
intricate balance of the rates of the shelving and the de-shelving significantly reduced ODMR contrast reported previously for
optical transitions for every defect. The exceptional variability of an ensemble. We identify an important indicator of ODMR, the
hBN optical rates, perhaps via strain36, may reflect the tunability presence of two bunching timescales in the g2(τ) measurements
of the defect energy levels in this 2D system if it can be controlled. which supports the idea that our variations in ODMR sign and
This could open routes to a room-temperature spin-photon contrast strength could be due to variations in photodynamics
interface where the spin readout can be reversibly and easily across defects, potentially caused by variations in strain.
tuned for use in sensing and memory-assisted quantum networks. ODMR-active defects possess a double peaked resonance with
Regarding the spin multiplicity and implications for the che- an average splitting of 35 MHz, consistent with a S = 1 state
mical structure of the defect, the yield of ODMR (~5%) and with a zero-field splitting on the order of 25 MHz. Angular-
prevalence of a doublet in our ODMR measurements (80%) is dependent measurements and simulations suggest that this
difficult to reconcile with a S = 12 system with a 1% abundance of continuum of values arises from a variation in the orientation
13C, although not impossible. Instead, we find that spin models of defects in the hBN plane. Further experimental and theore-
with S > 12 and low zero-field splitting parameters are viable tical work will be required to develop a deeper insight into the
alternatives. We find that all the ODMR data for our defects is microscopic structure and photophysics of these defects.
consistent with a S = 1 model with D = 25 MHz and E = 5 MHz, Regardless, these results reveal the potential for these defects as
by tuning the defect symmetry orientation in and out of the hBN a tuneable room-temperature spin-photon interface in a two-
plane. We consider that a range of defect orientations is highly dimensional material platform.
likely in this material, where the confocal scans and previous high
resolution TEM images of the same material show large regions
where the hBN layers are tilted relative to the substrate53. An Methods
alternative explanation could be that we are measuring a range of HBN Sheets. hBN was grown by metal organic vapour phase epitaxy (MOVPE) on
sapphire, as described in Mendelson et al47. Briefly, triethyl boron (TEB) and
different optical defects with low, but variable, D and E para- ammonia were used as boron and nitrogen sources with hydrogen used as a carrier
meters, dictated by local strain for example. While this is possible, gas. Growth was performed at low pressure (85 mBar) and at a temperature of
our analysis indicates that invoking different D and E parameters 1350 °C. Isolated defects and ensemble defects were generated by modifying the
is not necessary to model the data. Finally, while we demonstrate flow rate of TEB during growth, a parameter known to control the incorporation of
carbon within the resulting hBN sheets. For PL measurements, hBN sheets were
a spin triplet model is consistent with our data, a S ¼ 32 model is transferred to SiO2/Si substrates, using a water-assisted self-delamination process
difficult to distinguish from S = 1 in this field range and thus to avoid polymer contamination. Before measurements each device was treated in a
cannot be ruled out. UV/ozone cleaner for 15 min.
Experimental and theoretical reports indicate that the
structure of the defect is likely to contain carbon40,47,59–63.
Confocal microscopy. Optical measurements were carried out at room tem-
Defects in single crystalline hBN that show ODMR only under perature under ambient conditions using a home-built confocal microscopy
cryogenic conditions have been assigned to a spin-12 carbon setup. A continuous-wave 532-nm laser (Ventus 532, Laser Quantum) was sent
substitution defect (CB)40,56. This defect emits at 730 nm and through a 532 nm band-pass filter and focused on the device using an objective
the ODMR shows a broad 40 MHz resonance, attributed to lens with 100X magnification and a numerical aperture of 0.9. Control over
unresolved hyperfine coupling to neighbouring 11B and 14N excitation power was provided by an acousto-optic modulator (AA Optoelec-
tronics), with the first-order diffracted beam fibre-coupled into the confocal
nuclei56. It is difficult for us to conclude whether we are mea- setup. Two 550 nm long-pass filters (Thorlabs FEL550) were used to filter off
suring the same defect as these reports. While our ODMR reflected laser light from the collected emission, which was then sent either into
shares some features with those in ref. 40, such as defects that an avalanche photodiode (APD) (SPCM-AQRH-14-FC, Excelitas Technologies)
show positive and negative contrast, there are also interesting for recording photon count traces and observing the intensity of emission, or to
a CCD-coupled spectrometer (Acton Spectrograph, Princeton Instruments) via
differences: in addition to the difference in ZPL energy, the single-mode optical fibres (SM450 and SM600) for photoluminescence spec-
defects we study show ODMR at room temperature and the troscopy measurements. White-light images of the device were collected by
majority show a doublet, while the defects in ref. 40 do not show introducing the flippable mirror to divert the collection path to a CCD instead of
ODMR at room-temperature and no splitting was observed. the detection arm. This allowed easy device positioning and identification.
Intensity-correlation measurements were carried out using a Hanbury Brown
Most recently, a carbon-trimer structure has become a strong and Twiss interferometry setup using a 50:50 fibre beam-splitter and a time-to-
candidate for the single-photon emitting defects with ZPLs digital converter (quTAU, qutools) with 81-ps resolution.
~2 eV in hBN62. These defects are thermodynamically likely to
be formed63 and have been modelled to show ODMR with
highly variable contrast magnitude and sign, regulated by Optically detected magnetic resonance measurements. ODMR measurements
internal optical rates54, which is similar to our observations. were performed on the confocal setup described above. A 20-µm microstrip
microwave antenna was patterned photolithography over the hBN layers and
Therefore, we consider these structures, as well as larger carbon thermal evaporation of 100-nm Au on 20-nm Ti. The antenna was bonded to a
clusters predicted to be S = 163, as strong candidates for the coplanar waveguide on a printed circuit board (PCB), shorting the waveguide at
defect we measure. We note that the magnitude of zero-field the device. A 70-Hz square-wave modulation was applied to the microwave
splitting indicated by our data (<25 MHz) is small compared to amplitude to detect the change in PL counts as a function of microwave fre-
quency. A permanent magnet delivers an external static magnetic field in the
spin defects in diamond and SiC5,64 and organic molecules65. plane of the hBN surface and is changed in strength and orientation by dis-
However, we also note that 25 MHz corresponds to a spin-spin placing the magnet.

NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z

Data availability 28. Branny, A., Kumar, S., Proux, R. & Gerardot, B. D. Deterministic strain-
The datasets generated as part of the current study are available from the corresponding induced arrays of quantum emitters in a two-dimensional semiconductor.
authors upon reasonable request. Nat. Commun. 8, 15053 (2017).
29. Jungwirth, N. R. et al. Temperature dependence of wavelengths selectable
zero-phonon emission from single defects in hexagonal-boron nitride. Nano
Code availability Lett. 16, 6052–6057 (2016).
The codes used for the analysis included in the current study are available from the
30. Tran, T. T. et al. Robust multicolor single photon emission from point defects
corresponding authors upon reasonable request.
in hexagonal boron nitride. ACS Nano 10, 7331–7338 (2016).
31. Grosso, G. et al. Tunable and high-purity room temperature single-photon
Received: 25 June 2021; Accepted: 11 January 2022; emission from atomic defects in hexagonal boron nitride. Nat Commun. 8,
705 (2017).
32. Noh, G. et al. Stark tuning of single-photon emitters in hexagonal boron
nitride. Nano Lett. 18, 4710–4715 (2018).
33. Proscia, N. V. et al. Near-deterministic activation of room temperature
quantum emitters in hexagonal boron nitride. Optica 5, 1128–1134 (2018).
34. Nikolay, N. et al. Very large and reversible Stark-shift tuning of single emitters
References in layered hexagonal boron nitride. Phys. Rev. Appl. 11, 041001 (2019).
1. Awschalom, D. D., Hanson, R., Wrachtrup, J. & Zhou, B. B. Quantum
35. Gottscholl, A. et al. Initialization and read-out of intrinsic spin defects in a van
technologies with optically interfaced solid-state spins. Nat. Photonics 12,
der Waals crystal at room temperature. Nat. Mater. 19, 540–545 (2020).
516–527 (2018).
36. Ivády, V. et al. Ab initio theory of the negatively charged boron vacancy qubit
2. Atatüre, M., Englund, D., Vamivakas, N., Lee, S.-Y. & Wrachtrup, J. Material
in hexagonal boron nitride. npj Comput. Mater. 6, 41 (2020).
platforms for spin-based photonic quantum technologies. Nat. Rev. Mater. 3,
37. Gottscholl, A. et al. Room temperature coherent control of spin defects in
38–51 (2018).
hexagonal boron nitride. Science Advances 7, 14 (2021).
3. Gruber, A. et al. Scanning confocal optical microscopy and magnetic
38. Liu, W. et al. Rabi Oscillation of VB- spin in hexagonal boron nitride. Preprint
resonance on single defect centres. Science 276, 2012–2014 (1997).
at. https://arxiv.org/abs/2101.11220 (2021).
4. Acosta, V. & Hemmer, P. Nitrogen-vacancy centers: physics and applications.
39. Exarhos, A. L. et al. Magnetic-field-dependent quantum emission in hexagonal
MRS Bull. 38, 127–130 (2013).
boron nitride at room temperature. Nat. Commun. 10, 222 (2019).
5. Doherty, M. W. et al. The nitrogen-vacancy colour centre in diamond. Phys.
40. Chejanovsky, N. et al. Single-spin resonance in a van der Waals embedded
Rep. 528, 1–45 (2013).
paramagnetic defect. Nat. Mater. 20, 1079–1984 (2021).
6. Christle, D. J. et al. Isolated electron spins in silicon carbide with millisecond
41. Exarhos, A. L., Hopper, D. A., Grote, R. R., Alkauskas, A. & Bassett, L. C.
coherence times. Nat. Mater. 14, 160–163 (2014).
Optical signatures of quantum emitters in suspended hexagonal boron nitride.
7. Seo, H. et al. Quantum decoherence dynamics of divacancy spins in silicon
ACS Nano. 11, 3328–3336 (2017).
carbide. Nat. Commun. 7, 12935 (2016).
42. Shotan, Z. et al. Photoinduced modification of single-photon emitters in
8. Maurer, P. C. et al. Room-temperature quantum bit memory exceeding one
hexagonal boron nitride. ACS Photonics 3, 2490–2496 (2016).
second. Science 336, 1283–1286 (2012).
43. Khatri, P., Ramsay, A. J., Malein, R. N. E., Chong, H. M. H. & Luxmoore, I. J.
9. Robledo, L. et al. High-fidelity projective read-out of a solid-state spin
Optical gating of photoluminescence from colour centers in hexagonal boron
quantum register. Nature 477, 574–578 (2011).
nitride. Nano Lett. 20, 4256–4263 (2020).
10. Pla, J. J. et al. High-fidelity readout and control of a nuclear spin qubit in
44. Mendelson, N. et al. Engineering and tuning of quantum emitters in few-layer
silicon. Nature 496, 334 (2013).
hexagonal boron nitride. ACS Nano 13, 3132–3140 (2019).
11. Childress, L. et al. Coherent dynamics of coupled electron and nuclear spin
45. Abidi, I. H. et al. Selective defect formation in hexagonal boron nitride. Adv.
qubits in diamond. Science.13, 314 (2006).
Opt. Mater. 7, 1900397 (2019).
12. Bradley, C. E. et al. A ten-qubit solid-state spin register with quantum
46. Stern, H. L. et al. Spectrally resolved photodynamics of individual emitters in large-
memory up to one minute. Phys. Rev. X 9, 3 (2019).
area monolayers of hexagonal boron nitride. ACS Nano 13, 4538–4547 (2019).
13. Sukachev, D. D. et al. Silicon-vacancy spin qubit in diamond: a quantum
47. Mendelson, N. et al. Identifying carbon as the source of visible single-photon
memory exceeding 10 ms with single-shot state readout. Phys. Rev. Lett. 119,
emission from hexagonal boron nitride. Nat. Mater. 20, 321–328 (2021).
223602 (2017).
48. Vogl, T., Campbell, G., Buchler, B. C., Lu, Y. & Lam, P. K. Fabrication and
14. Metsch, M. H. et al. Initialization and readout of nuclear spins via a negatively
deterministic transfer of high-quality quantum emitters in hexagonal boron
charged silicon-vacancy centre in diamond. Phys. Rev. Lett. 122, 190503
nitride. ACS Photonics 5, 2305–2312 (2018).
(2019).
49. Dietrich, A., Doherty, M. W., Aharonovich, I. & Kubanek, A. Solid-state single
15. Aharonovich, I., Englund, D. & Toth, M. Solid-state single-photon emitters.
photon source with Fourier transform limited lines at room temperature.
Nature Photon 10, 631–641 (2016).
Phys. Rev. B 101, 081401 (2020).
16. Sipahigil, A. et al. An integrated diamond nanophotonics platform for
50. Vogl, T., Lecamwasam, R., Buchler, B. C., Lu, Y. & Lam, P. K. Compact cavity-
quantum-optical networks. Nature 354, 6314 (2016).
enhanced single-photon generation with hexagonal boron nitride. ACS
17. Evans, R. E. et al. Photon-mediated interactions between quantum emitters in
Photonics 6, 1955–1962 (2019).
a diamond nanocavity. Science 362, 662–665 (2018).
51. Proscia, N. V. et al. Microcavity-coupled emitters in hexagonal boron nitride.
18. Taminiau, T. H., Cramer, J., van der Sar, T., Dobrovitski, V. V. & Hanson, R.
Nanophotonics 9, 2937–2944 (2020).
Universal control and error correction in multi-qubit spin registers in
52. Froch, J. et al. Coupling hexagonal boron nitride quantum emitters to
diamond. Nat. Nanotechnol. 9, 171–176 (2014).
photonic crystal cavities. ACS Nano 14, 7085–7091 (2020).
19. Maze, J. et al. Nanoscale magnetic sensing with an individual electronic spin in
53. Chugh et al. Flow modulation epitaxy of hexagonal boron nitride. 2D Mater.
diamond. Nature 455, 644–647 (2008).
5, 045018 (2018).
20. Schmitt et al. Submillihertz magnetic spectroscopy performed with a
54. Golami, O. et al. Ab initio and group theoretical study of properties of the C2CN
nanoscale quantum sensor. Science 356, 6340 (2017).
carbon trimer in hBN. Preprint at. https://arxiv.org/abs/2110.09628 (2021).
21. Srivastava, A. et al. Optically active quantum dots in monolayer WSe2. Nat.
55. Fanciulli, M. Electron paramagnetic resonance and relaxation in BN and
Nanotechnol. 10, 491–496 (2015).
BN:C. Philos. Mag. 76, 3 (1997).
22. He, Y. M. et al. Single quantum emitters in monolayer semiconductors. Nat.
56. Katzir, A., Suss, J. T., Zunger, A. & Halperin, A. Point defects in hexagonal
Nanotechnol. 10, 497–502 (2015).
boron nitride. I. EPR, thermoluminescence, and thermally-stimulated-current
23. Koperski, M. et al. Single photon emitters in exfoliated WSe2 structures. Nat.
measurements. Phys. Rev. B. 11, 6 (1975).
Nanotechnol. 10, 503–506 (2015).
57. Auburger, P. & Gali, A. Towards ab initio identification of paramagnetic
24. Chakraborty, C., Kinnischtzke, L., Goodfellow, K. M., Beams, R. & Vamivakas,
substitutional carbon defects in hexagonal boron nitride acting as quantum
A. N. Voltage-controlled quantum light from an atomically thin
bits. Phys. Rev. B. 104, 075410 (2021).
semiconductor. Nat. Nanotechnol. 10, 507–511 (2015).
58. Carter et al. Spin coherence and echo modulation of the silicon vacancy in
25. Tonndorf, P. et al. Single-photon emission from localized excitons in an
4H- SiC at room temperature. Phys. Rev. B. 92, 161202 (2015).
atomically thin semiconductor. Optica 2, 347–352 (2015).
59. Tawfik, S. A. et al. First-principles investigation of quantum emission from
26. Tran, T. T., Bray, K., Ford, M. J., Toth, M. & Aharonovich, I. Quantum
hBN defects. Nanoscale 9, 13575–13582 (2017).
emission from hexagonal boron nitride monolayers. Nat. Nanotechnol. 11,
60. Abdi, M. et al. Color centers in hexagonal boron nitride monolayers: a group
37–41 (2016).
theory and ab initio analysis. ACS Photonics 5, 1967–1976 (2018).
27. Palacios-Berraquero, C. et al. Large-scale quantum-emitter arrays in
61. Sajid, A. et al. VNCB defect as source of single photon emission from
atomically thin semiconductors. Nat Commun. 8, 15093 (2017).
hexagonal boron nitride. 2D Mater. 7, 031007 (2020).

8 NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-28169-z ARTICLE

62. Jara et al. First-principles identification of single photon emitters based on carbon Additional information
clusters in hexagonal boron nitride. J. Phys. Chem. A 125, 1325–1335 (2021). Supplementary information The online version contains supplementary material
63. Maciaszek et al. Thermodynamics of carbon point defects in hexagonal boron available at https://doi.org/10.1038/s41467-022-28169-z.
nitride. Preprint at. https://arxiv.org/abs/2110.12167 (2021).
64. Koehl, W. et al. Room temperature coherent control of defect spin qubits in Correspondence and requests for materials should be addressed to Hannah L. Stern or
silicon carbide. Nature 479, 84–87 (2011). Mete Atatüre.
65. Yang, T. C. et al. Zero-field magnetic resonance of the photo-excited triplet
state of pentacene at room temperature. J. Chem. Phys. 113, 11194 (2000). Peer review information Nature Communications thanks the anonymous reviewers for
their contribution to the peer review of this work. Peer reviewer reports are available.

Acknowledgements Reprints and permission information is available at http://www.nature.com/reprints

The authors would like to thank L. Bassett, A. Alkauskas and S. Stranks for useful
discussions. We thank the research group of Sam Stranks for use of their laser for the Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
EPR measurements. This work is funded by ERC Proof-of-Concept grant PEGASOS published maps and institutional affiliations.
(862405), ERC Advanced Grant PEDESTAL (884745), Australian Research Council
(CE200100010), and the Asian Office of Aerospace Research & Development (FA2386-
20-1-4014). H.L.S is funded by Trinity College, Cambridge. S.E.B. acknowledges funding Open Access This article is licensed under a Creative Commons
from the EPSRC CDT in Nanoscience and Nanotechnology (NanoDTC, Grant No. EP/ Attribution 4.0 International License, which permits use, sharing,
S022953/1). Q.G. acknowledges financial support by the China Scholarship Council and adaptation, distribution and reproduction in any medium or format, as long as you give
the Cambridge Commonwealth, European & International Trust. S.S. acknowledges appropriate credit to the original author(s) and the source, provide a link to the Creative
funding from the ERC Synergy Grant SC2 No. 610115. Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
Author contributions indicated otherwise in a credit line to the material. If material is not included in the
M.A. and H.L.S. conceived the project. H.L.S., Q.G., J.J., S.E.B. and S.S. performed article’s Creative Commons license and your intended use is not permitted by statutory
experiments and carried out analysis. N.M., D.C., H.H.T. and I.A. provided the samples. regulation or exceeds the permitted use, you will need to obtain permission directly from
M.A., H.L.S., Q.G., J.J, S.E.B., N.M., S.S., H.S. and I.A. contributed to discussion of the the copyright holder. To view a copy of this license, visit http://creativecommons.org/
results and the preparation of the manuscript. licenses/by/4.0/.

Competing interests © The Author(s) 2022

The authors declare no competing interests.

NATURE COMMUNICATIONS | (2022)13:618 | https://doi.org/10.1038/s41467-022-28169-z | www.nature.com/naturecommunications 9