LETTERS

PUBLISHED ONLINE: 6 MARCH 2017 | DOI: 10.1038/NNANO.2017.18

Atomic-scale sensing of the magnetic dipolar eld

from single atoms
Taeyoung Choi1, William Paul1, Steffen Rolf-Pissarczyk2,3, Andrew J. Macdonald4,
Fabian D. Natterer1,5, Kai Yang1,6, Philip Willke1,7, Christopher P. Lutz1* and Andreas J. Heinrich8,9*

Spin resonance provides the high-energy resolution needed to
determine biological and material structures by sensing weak
magnetic interactions1. In recent years, there have been
notable achievements in detecting2 and coherently control-
ling37 individual atomic-scale spin centres for sensitive local
magnetometry810. However, positioning the spin sensor and
characterizing spinspin interactions with sub-nanometre pre-
cision have remained outstanding challenges11,12. Here, we use
individual Fe atoms as an electron spin resonance (ESR)
sensor in a scanning tunnelling microscope to measure the
magnetic eld emanating from nearby spins with atomic-
scale precision. On articially built assemblies of magnetic
atoms (Fe and Co) on a magnesium oxide surface, we
measure that the interaction energy between the ESR sensor
and an adatom shows an inverse-cube distance dependence
(r 3.010.04). This demonstrates that the atoms are predomi-
nantly coupled by the magnetic dipoledipole interaction,
which, according to our observations, dominates for atom sep-
arations greater than 1 nm. This dipolar sensor can determine
the magnetic moments of individual adatoms with high accu-
racy. The achieved atomic-scale spatial resolution in remote
sensing of spins may ultimately allow the structural imaging
of individual magnetic molecules, nanostructures and
spin-labelled biomolecules.
In traditional magnetic resonance experiments, electromagnetic
radiation interacts resonantly with an ensemble of electron or
nuclear spins via absorption and emission of photons. This typically
requires an ensemble of 107 to 1010 spins1. Detecting individual
spins and their interactions has been pursued by means of
atomic-scale spin centres such as optically trapped ions4, nitrogen
vacancy centres in diamond810, spins in quantum dots5, dopant
atoms in semiconductors6 and single-molecule magnets in break
junctions7. A resolution of 10 nm has been achieved for the con-
trolled positioning and spin-resonant imaging of spin centres in
solid-state materials9,11, while atomic-scale control remains a
highly challenging goal12.
The scanning tunnelling microscope (STM) excels at positioning
individual spin centres with atomic precision, and has been fruitful
for studying the magnetic states of atoms and molecules1319. In
magnetic materials, the magnetic dipoledipole interaction
between nearby electron spins is usually eclipsed by much larger
interactions such as RudermanKittelKasuyaYosida (RKKY)20
or superexchange14,21. On the other hand, the dipoledipole inter-
action dominates for spins in an insulator separated by large
distances. This interaction is essential for structural imaging of
molecules but its use is currently limited to ensemble measure-
ments1. In STM, however, the dipoledipole coupling energies of
individual spin centres are too small to detect by tunnelling spec-
troscopy, even at low temperature. Here, we combine the superior
energy resolution provided by ESR (also known as electron para-
magnetic resonance, EPR) with the capabilities of STM (ref. 22),
to measure magnetic dipoledipole interactions in precisely con-
trolled atomic arrangements of magnetic atoms on a surface, a
major step towards structural imaging of individual molecules.
Figure 1a shows a schematic of our ESRSTM experiment oper-
ating at low temperature (0.61.2 K). We deposited individual Fe
and Co atoms onto bilayer magnesium oxide (MgO) lms grown
on a silver (Ag) substrate (Fig. 1b). The MgO lm reduces the scat-
tering between the spin of adatoms and the Ag substrate electrons,
leading to long spin relaxation times23. Both Fe (spin S = 2) and Co
(S = 3/2) adsorb atop oxygen on MgO, which results in a strong
easy-axis magnetic anisotropy perpendicular to the surface
plane22,24,25. The two lowest energy quantum spin states of the Fe
atoms on MgO (|, |) are oriented in two opposite out-of-
plane directions and are split in energy by an out-of-plane magnetic
eld (Bz)22,24. This Zeeman splitting is measured directly using a
radiofrequency (RF) electric eld applied in the tipsample junc-
tion22,26. When the frequency of the continuously applied RF electric
eld is resonant with the Zeeman splitting of an Fe atom, the eld
drives spin transitions between the two states, depopulating the
ground state from its thermal equilibrium. This change of the state
population is detected with a spin-polarized STM tip by tunnelling
magnetoresistance15,22,27, producing the pronounced ESR signal
from individual Fe atoms shown in Fig. 2a (Supplementary Section 2).
We rst focus on the magnetic interaction between two Fe atoms.
Using atom manipulation, the separations of these atoms were
precisely set to increments of the MgO lattice (Supplementary
Section 1). We positioned the STM tip above one of the Fe atoms
(sensor atom) in the pair and measured its ESR signal, which
showed two spectral features (Fig. 2a) corresponding to two
oppositely oriented magnetic states, | and | of the target Fe
atom (Fig. 2c). The ESR frequency difference f, which gives the
magnetic interaction energy, grows as the atoms are moved closer
in distance (Fig. 2b, red squares). This measurement was performed
on more than ten different atom pairs (Supplementary Section 4)
and yielded reproducible interaction energies independent of
the detailed structure of the tip apex, temperature and applied
magnetic eld (Supplementary Section 5). For FeFe distances

1
IBM Almaden Research Center, San Jose, California 95120, USA. 2 Max Planck Institute for the Structure and Dynamics of Matter, Hamburg 22761, Germany.
3
Max Planck Institute for Solid State Research, Stuttgart 70569, Germany. 4 University of British Columbia & Quantum Matter Institute, Vancouver, BC V6T 1Z4,
Canada. 5 Institute of Physics, cole Polytechnique Fdrale de Lausanne, CH-1015 Lausanne, Switzerland. 6 School of Physical Sciences and Key Laboratory of
Vacuum Physics, University of Chinese Academy of Sciences, Beijing 100049, China. 7 IV. Physical Institute, University of Gttingen, Friedrich-Hund-Platz 1,
37077 Gttingen, Germany. 8 Center for Quantum Nanoscience, Institute for Basic Science (IBS), Seoul 03760, Republic of Korea. 9 Physics Department,
Ewha Womans University, Seoul, Republic of Korea. Present address: Center for Quantum Nanoscience, Institute for Basic Science (IBS), Seoul 03760,
Republic of Korea; Department of Physics, Ewha Womans University, Seoul 03760, Republic of Korea. * e-mail: cplutz@us.ibm.com; heinrich@ewha.ac.kr

420 NATURE NANOTECHNOLOGY | VOL 12 | MAY 2017 | www.nature.com/naturenanotechnology

2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2017.18 LETTERS
a VDC b
VDC+RF
VRF

SP-STM
tip

Bz

Bx
Sensor
Target
MgO
1 nm
Ag(1
00)

Figure 1 | Experimental set-up for ESRSTM. a, Schematic for sensing the dipoledipole interaction between a sensor atom (Fe) and a target atom. Electron
spin resonance of individual Fe atoms on bilayer MgO/Ag(100) is driven by a radiofrequency electric eld at the tunnelling junction due to VRF. The
out-of-plane magnetic eld Bz sets the resonance frequency. An in-plane magnetic eld Bx mixes the spin states of Fe to increase the ESR signal22.
A spin-polarized (SP) STM tip measures the spin resonance signal via tunnelling magnetoresistance. The target atom creates a local magnetic eld
(dipolar eld) that is detected by the sensor atom (Fe). b, Constant-current STM image of four Fe atoms (0.17 nm apparent height) on the surface.
Imaging conditions are VDC = 0.1 V, I = 10 pA, Bz = 0.18 T, Bx = 5.7 T and T = 1.2 K.

a b
0 FeFe
r3 fit
ESR signal I (fA)

5 FeCo
r3 fit

10 1 Fetall Fe
r3 fit
f
15
23.0 23.1 23.2 23.3
f (GHz)

Frequency (GHz)

c Dipole coupled 0.1

Sensor only
2Edd/h f

f0 f

2Edd/h
Sensor 0.01
atom 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Sensor Target
atom atom Distance, r (nm)

Figure 2 | Magnetic dipoledipole interaction. a, ESR spectrum (black curve) of an Fe atom when another Fe atom (target) is positioned 2.46 nm away
(Bz = 0.15 T, T = 1.2 K, VDC = 5 mV, IDC = 1 pA, VRF = 10 mVpp). Two Lorentzian ts (red curve) yield the frequency splitting (f ). The two spectral peaks
correspond to the two magnetic states of the target atom because each state of the target atom generates a different local magnetic eld at the sensor
atom. b, Measured splitting as a function of distance r of atom pairs of FeFe (red squares), FeCo (blue circles) and Fetall Fe (green diamonds). The data
are tted to f r for each type of target atom (Supplementary Fig. 5). The slopes give = 3.01 0.04 (FeFe), 2.98 0.04 (FeCo) and 2.94 0.08
(Fetall Fe), indicating a magnetic dipoledipole interaction in each case. The straight solid lines are ts to equation (2), using a xed exponent of = 3 to
extract magnetic moments for each type of target atom. The magnetic moments are 5.44 0.03 (Fe), 5.88 0.06 (Co) and 4.35 0.08 (tall Fe). The
short-range data (r < 1 nm, shaded region) were excluded in the tting process (Supplementary Fig. 6). Error bars in the x and y axes represent uncertainty
in the distance of atom pairs and the measured f error due to frequency drift, respectively (Supplementary Sections 1 and 4). c, Schematic of the energy
levels of the sensor atom, modied by its dipoledipole interaction with the target. The resonant frequency of the sensor atom ( f0) is split into two
frequencies ( f and f) corresponding to the two spin states of the target atom. The two grey arrows represent the respective ESR frequencies in a, and the
difference f = f f is the measured splitting. Edd is the magnetic dipole-dipole energy.

NATURE NANOTECHNOLOGY | VOL 12 | MAY 2017 | www.nature.com/naturenanotechnology 421

2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
LETTERS
greater than 1.0 nm, a t of the measured splitting to a power-law
relation f r yields 3.01 0.04, in excellent agreement with the
exponent = 3 characteristic of the magnetic dipoledipole
interaction (Supplementary Fig. 5). We found that for atoms separ-
ated by less than 1 nm (grey region in Fig. 2b), the coupling deviates
markedly from a pure dipoledipole interaction, indicating that it is
inuenced by additional interactions such as exchange coupling
(Supplementary Section 4.3).
The magnetic dipoledipole energy between two moments is
given in general by
0 1 
(1) 
(1) 
(2) 
Edd =
m
m
(2) 3 m
r

r m
4 r 3
where 0 is the vacuum permeability, r is the separation between the
magnetic moments, r is a unit vector along the line linking the
moments, and m
(i) is the magnetic moment of atom i. For Fe on
MgO, the magnetic moments of the Fe atoms are fully polarized
in the out-of-plane direction (Fig. 2c) due to the large magnetic ani-
sotropy22,24. As a result, the canting of the moment due to the in-
plane magnetic eld can be neglected (Supplementary Section 4.1),
and the dipoledipole interaction can be simplied to
0 1 
(1) (2) 
Edd = mz mz
4 r3
The frequency difference of the two ESR peaks (or dips), plotted in
Fig. 2b, becomes f = f f = 4Edd/h, which is four times the mag-
netic dipoledipole energy (Fig. 2c). The measured dipolar inter-
action was used to extract the magnetic moment of the Fe atoms.
A single parameter t to equation (2), using a xed exponent of
= 3 (red line in Fig. 2b), yields a magnetic moment for Fe of
m(Fe)
z = 5.44 0.03B . ESRSTM thus provides a new means to
measure the magnetic moment of individual atoms with high pre-
cision. The measured moment of Fe agrees with previous modelling
results (5.2B for multiplet modelling guided by X-ray studies24) and
it approaches that of a free Fe atom (6B), indicating that the orbital
angular momentum is largely preserved24. Our ESR spectral line-
width of 10 MHz (Fig. 2b) allows us to resolve the distance to a
target atom to within one atomic lattice site for separations up to
4 nm, a resolution necessary for atomic-scale sensing of
individual spins.
The relative amplitude of the two ESR peaks gives information
about the thermal occupation of the two magnetic states of the
target atom, which are separated by the Zeeman energy of 0.1 meV
(23 GHz) for the out-of-plane magnetic eld Bz 0.15 T used
here. The larger ESR peak corresponds to the ground state (|)
of the target atom, which is occupied with greater probability
(Fig. 2c). The ratio of the two ESR peak heights (0.36 0.05) in
Fig. 2a agrees well with the Boltzmann factor (0.4 at T = 1.2 K)
(Supplementary Section 6). The spin relaxation time of the Fe
target atom was measured by pumpprobe spectroscopy to be
100 s (refs 22 and 23), which is much faster than the data point
averaging time in the ESR spectrum (1 s). Therefore, the ESR spec-
trum on the Fe sensor statistically samples the thermal population
of the target Fe spin at each data point.
The accurate determination of the Fe magnetic moment enables us
to use the Fe atom as a sensor for measuring the magnetic moment of
other target atoms. More specically, because the sensor Fe moment
is perpendicular to the surface plane, equation (2) allows us to obtain
the out-of-plane moment of the target atom (Supplementary
Section 4.1). For FeCo pairs (blue circles in Fig. 2b), a t to the
power-law relation f r yields = 2.98 0.04, showing that Fe
and Co atoms also couple through a magnetic dipoledipole inter-
action for r > 1 nm. Using a single parameter t to equation (2)
(blue line in Fig. 2b) we extract the out-of-plane component of the
422
2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2017.18
Co magnetic moment to be m(Co) z = 5.88 0.06B , which is again
close to free Co (6B)25. We note that the accuracy of our new tech-
nique, 1%, is unprecedented for determining magnetic moments
of spins on surfaces and should allow stringent tests of theoretical
and computational models.
We observed that after some atom manipulation attempts
(10% yield), Fe atoms appeared about 0.06 nm taller than the
common as-deposited Fe atoms (height of 0.17 nm). We refer to
this species as tall Fe. The tall Fe atoms also sit atop oxygen on
MgO and can be reversibly switched back to the as-deposited Fe
species by applying a voltage pulse (Supplementary Section 1)28.
Following the previous analysis, for dimers containing a tall Fe
(1)
atom, we obtained a power law exponent of 2.94 0.08 and a
magnetic moment of 4.35 0.07B, indicating that the Fe sensor
can indeed be used to measure the magnetic moment of otherwise
uncharacterized species. The tall Fe and as-deposited Fe may be
associated with a different charge state of the atoms28, which
inuences the electron conguration and may result in the difference
in the magnetic moments.
To put these results into perspective, we note that the magnetic
moments of Fe and Co atoms on the MgO surface were determined
in recent experiments to be 5.2B and 5.4B by multiplet simulations
of X-ray magnetic circular dichroism (XMCD) spectra24,25.
These values are similar to the ones obtained in our ESRSTM
(2)
experiments. In contrast to ensemble measurements in XMCD
(>1010 atoms)24,25, the present work addresses individual magnetic
species, which offers the ability to distinguish between atoms on
different binding sites or with different charge states.
The symmetry of a structure is closely connected to the degener-
acy of its quantum states. In the following, we use a single sensor atom
to probe such symmetry of the dipolar eld from several target atoms.
We used STM atom manipulation to assemble four Fe atoms into a
square with exactly ten atomic lattice sites (2.877 nm) on a side. First,
we placed a fth (sensor) Fe atom off-centre in the square to create
different distances to each target atom. There are 24 = 16 possible
states of the target atoms, and each yields a distinct ESR frequency
on the sensor Fe (vertical lines in Fig. 3). The spectrum shows a domi-
nant peak (Fig. 3a), due to target atoms residing most often in their
ground states, according to their Boltzmann distribution. For the
purpose of describing the spectrum due to multiple target atoms,
the interactions between pairs of target atoms can be neglected
because they are much smaller than the Zeeman energies of each
atom. The sensor atom therefore probes the sum of the magnetic
elds created by all target atoms (Supplementary Section 7).
Next, we introduced higher symmetry into the structure of
Fig. 3a by moving the sensor Fe atom to the exact centre of the
square (Fig. 3b). Because in this case the separation of the sensor
and the target atoms is equal, many of the ESR peaks of Fig. 3a
become degenerate (thick vertical lines, Fig. 3b). The intensity of
the ESR lines follows the degeneracy of the corner atom states
(1:4:6:4:1), weighted by the corresponding Boltzmann distribution.
For example, the peak at f 0.183 GHz (pink, Fig. 3b) arises
from the four degenerate cases in which exactly one corner atom
is in its excited state. At 1.2 K, the population of excited states
increases, transferring spectral weight to higher-frequency peaks,
which correspond to progressively more of the corner atoms
being in their excited state. Furthermore, this spin structure can
be used as a local thermometer by tting the ESR spectra to the pair-
wise dipoledipole interaction. We nd temperatures from the ESR
ts of Tt = 0.72 0.02 K and 1.43 0.13 K for Fig. 3b, which agree
well with the temperatures measured by a thermometer on the STM.
The magnetic sensing provided by spin-resonant Fe atoms also
offers a means to perform magnetic imaging in the STM. We illus-
trate this by remotely sensing the position and moment of an
unknown target spin centre. We employ trilateration, the
geometrical technique used for macroscopic location by the
NATURE NANOTECHNOLOGY | VOL 12 | MAY 2017 | www.nature.com/naturenanotechnology
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2017.18 LETTERS
a

T = 0.6 K

30
T = 1.2 K

I (fA)
60

90

1 nm 120
0.0 0.2 0.4 0.6 0.8 1.0
f (GHz)

T = 0.6 K
0

30
T = 1.2 K
I (fA)

60

90

1 nm 120
0.0 0.2 0.4 0.6 0.8 1.0
f (GHz)

Figure 3 | Control of state degeneracy in an engineered nanostructure. a, Five Fe atom structure with four equally spaced corner atoms and a middle Fe
(sensor atom) that is positioned at (3, 1) lattice sites from the centre of the square (oxygen sites indicated by white open circles). The grey line indicates
the ESR frequency of the Fe sensor with all corner (target) atoms in their magnetic ground state. Pink lines indicate ESR frequencies with one of the four
target atoms in the excited state. Additional ESR lines arise (with less intensity) when two or more of the corner atoms are in the excited state. No ESR
transitions are degenerate. b, The Fe sensor atom is placed in the exact centre of the square at 2.034 nm from the four target atoms. This creates a four-fold
symmetry, which leads to degeneracy (1:4:6:4:1) of the excited spin states. For example, the four pink lines in a are now degenerate at f 0.183 GHz. The
ESR signals of a and b were taken at two different sample temperatures: 0.6 K (upper traces) and 1.2 K (lower) measured by a thermometer. The red and
blue solid curves of a and b are modelled based on the dipole energy, the degeneracy of the states and the thermal population of the states. For clarity, the
1.2 K spectra are offset by 60 fA. Imaging conditions: VDC = 10 mV, I = 10 pA, Bz = 0.15 T, T = 1.2 K.

a Sensor 1 b c
0 0.0 0
Sensor 1 0.2
30 Calculated spectrum 30 Sensor 1
Deconvoluted signal (fA)

excluding middle atom 0.4 Lorentzian fit

60 60
0 0
I (fA)

Unknown
30 Sensor 2 30 Sensor 2
Lorentzian fit
60 60
Sensor 3 Sensor 2
0 0

30 Sensor 3 30 Sensor 3
Lorentzian fit
1 nm 60 60
0.0 0.2 0.4 0.0 0.2 0.4
f (GHz) f (GHz)

Figure 4 | Magnetic imaging by using trilateration. a, An STM image of the ve-atom structure (the same image as the Fig. 3a). We measure ESR of each
sensor atom to identify the magnetic moment and location of the middle (assumed to be unknown) atom. b, Measured ESR spectra (T = 0.6 K) from each
sensor atom (black) and the predicted unbroadened spectrum (red) due to all atoms excluding the unknown. c, ESR spectra of b after deconvolving the
effect of the other atoms. The results show ESR spectra (peaks are indicated by black arrows) due solely to dipolar interaction of the unknown atom and
each sensor. An intersection point of the three coloured circles obtained from c is indicated by the black circle in a, which gives a resolution of the unknown
that is better than 0.1B and 0.1 nm.

NATURE NANOTECHNOLOGY | VOL 12 | MAY 2017 | www.nature.com/naturenanotechnology 423

2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
LETTERS
Global Positioning System (GPS). In this nano-GPS scheme, the
distance to several precisely known reference points, each an Fe
sensor atom, yields a unique location for the unknown spin. It
falls at the intersection of the circles of known radius around each
sensor. Because the targets magnetic moment is also considered
to be unknown, we solve for it numerically by nding the
moments value for which the circles converge at a unique point
in space. This is illustrated in the ve-atom structure of Fig. 4a
(the same structure as in Fig. 3a), where we treat the middle Fe
atom as an unknown spin centre and sense its location and
moment purely from the ESR spectra of three Fe sensor atoms
(Fig. 4b). As a slight complication, the ESR signal from each
sensor atom is convolved by the effect of the other nearby sensor
atoms. As the effect of the other Fe atoms can be predicted with
high accuracy (red curve in Fig. 4b), we remove their effect on the
spectra by numerical deconvolution. The result (Fig. 4c) shows
derived spectra due only to interaction of the middle (target) atom
with each individual sensor atom in turn. Trilateration to determine
the unknown atoms properties from these derived spectra gives a
resolution that is better than 0.1 nm (indicated by the black circle
in Fig. 4a), and better than 0.1B (Supplementary Section 8).
The Fe sensor complements the unique atom manipulation
strengths of STM with the high energy resolution of ESR to detect
magnetic dipolar elds. At present, the detection range (4 nm)
of the Fe sensor is mainly limited by the spectral linewidth of our
ESR signal. This can be narrowed by reducing the applied RF ampli-
tude and the effect of mechanical noise of the set-up22, and possibly
by lengthening the spin coherence time by employing thicker MgO
layers. The nano-GPS scheme presented here provides a new route
to atomic-scale structural imaging of nanostructures that can be
assembled or placed on the surface, such as magnetic molecules
and spin-labelled biomolecules.
Data availability. The authors declare that the main data supporting
the ndings of this study are available within the Letter and its
Supplementary Information. Extra data or materials are available
from T. Choi (taeng0@gmail.com) upon reasonable request.
Received 5 August 2016; accepted 26 January 2017;
published online 6 March 2017
References
1. Abragam, A. & Bleaney, B. Electron Paramagnetic Resonance of Transition Ions
(Oxford Univ. Press, 2012).
2. Rugar, D., Budakian, R., Mamin, H. J. & Chui, B. W. Single spin detection by
magnetic resonance force microscopy. Nature 430, 329332 (2004).
3. Nowack, K. C., Koppens, F. H. L., Nazarov, Y. V. & Vandersypen, L. M. K.
Coherent control of a single electron spin with electric elds. Science 318,
14301433 (2007).
4. Kotler, S., Akerman, N., Navon, N., Glickman, Y. & Ozeri, R. Measurement of
the magnetic interaction between two bound electrons of two separate ions.
Nature 510, 376380 (2014).
5. Koppens, F. H. L. et al. Driven coherent oscillations of a single electron spin in a
quantum dot. Nature 442, 766771 (2006).
6. Pla, J. J. et al. High-delity readout and control of a nuclear spin qubit in silicon.
Nature 496, 334338 (2013).
7. Thiele, S. et al. Electrically driven nuclear spin resonance in single-molecule
magnets. Science 344, 11351138 (2014).
8. Grinolds, M. S. et al. Nanoscale magnetic imaging of a single electron spin under
ambient conditions. Nat. Phys. 9, 215219 (2013).
424
2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE NANOTECHNOLOGY DOI: 10.1038/NNANO.2017.18
9. Mamin, H. J. et al. Nanoscale nuclear magnetic resonance with a nitrogen-
vacancy spin sensor. Science 339, 557560 (2013).
10. Staudacher, T. et al. Nuclear magnetic resonance spectroscopy on a (5-
nanometer)3 sample volume. Science 339, 561563 (2013).
11. Toyli, D. M., Weis, C. D., Fuchs, G. D., Schenkel, T. & Awschalom, D. D. Chip-
scale nanofabrication of single spins and spin arrays in diamond. Nano Lett 10,
31683172 (2010).
12. Fuechsle, M. et al. A single-atom transistor. Nat. Nanotech. 7, 242246 (2012).
13. Heinrich, A. J., Gupta, J. A., Lutz, C. P. & Eigler, D. M. Single-atom spin-ip
spectroscopy. Science 306, 466469 (2004).
14. Hirjibehedin, C. F., Lutz, C. P. & Heinrich, A. J. Spin coupling in engineered
atomic structures. Science 312, 10211024 (2006).
15. Wiesendanger, R. Spin mapping at the nanoscale and atomic scale. Rev. Mod.
Phys. 81, 14951550 (2009).
16. Heinrich, B. W., Braun, L., Pascual, J. I. & Franke, K. J. Protection of excited spin
states by a superconducting energy gap. Nat. Phys. 9, 765768 (2013).
17. Tsukahara, N. et al. Adsorption-induced switching of magnetic anisotropy in a
single iron(II) phthalocyanine molecule on an oxidized Cu(110) surface. Phys.
Rev. Lett. 102, 167203 (2009).
18. Khajetoorians, A. A. et al. Atom-by-atom engineering and magnetometry of
tailored nanomagnets. Nat. Phys. 8, 497503 (2012).
19. Jacobson, P. et al. Quantum engineering of spin and anisotropy in magnetic
molecular junctions. Nat. Commun. 6, 8536 (2015).
20. Zhou, L. et al. Strength and directionality of surface RudermanKittelKasuya
Yosida interaction mapped on the atomic scale. Nat. Phys. 6, 187191 (2010).
21. Chen, X. et al. Probing superexchange interaction in molecular magnets by spin-
ip spectroscopy and microscopy. Phys. Rev. Lett. 101, 197208 (2008).
22. Baumann, S. et al. Electron paramagnetic resonance of individual atoms on a
surface. Science 350, 417420 (2015).
23. Paul, W. et al. Control of the millisecond spin lifetime of an electrically probed
atom. Nat. Phys. http://dx.doi.org/10.1038/nphys3965 (2016).
24. Baumann, S. et al. Origin of perpendicular magnetic anisotropy and large orbital
moment in Fe atoms on MgO. Phys. Rev. Lett. 115, 237202 (2015).
25. Rau, I. G. et al. Reaching the magnetic anisotropy limit of a 3d metal atom.
Science 344, 988992 (2014).
26. Paul, W., Baumann, S., Lutz, C. P. & Heinrich, A. J. Generation of constant-
amplitude radio-frequency sweeps at a tunnel junction for spin resonance STM.
Rev. Sci. Instrum. 87, 074703 (2016).
27. Julliere, M. Tunneling between ferromagnetic lms. Phys. Lett. A 54,
225226 (1975).
28. Repp, J., Meyer, G., Olsson, F. E. & Persson, M. Controlling the charge state of
individual gold adatoms. Science 305, 493495 (2004).
Acknowledgements
The authors thank B. Melior for expert technical assistance and T. Greber for fruitful
discussions. We gratefully acknowledge nancial support from the Ofce of Naval Research.
W.P. thanks the Natural Sciences and Engineering Research Council of Canada (NSERC) for
fellowship support. A.J.M. acknowledges nancial support from the NSERC CREATE and
PGS D programmes. F.D.N. appreciates nancial support from the Swiss National Science
Foundation (P300P2_158468 and PZ00P2_167965). K.Y. thanks the National Natural
Science Foundation of China grant no. 61471337) for nancial support. P.W. and S.R.
gratefully acknowledge nancial support from the German academic exchange service.
Author contributions
T.C., W.P., C.P.L. and A.J.H. conceived the projects. T.C., W.P., S.R., A.J.M. and F.D.N.
performed the experiments and analysed the data. T.C. wrote the manuscript and
developed theoretical models and simulations. All authors discussed the results and
commented on the manuscript.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence and
requests for materials should be addressed to C.P.L. and A.J.H.
Competing nancial interests
The authors declare no competing nancial interests.
NATURE NANOTECHNOLOGY | VOL 12 | MAY 2017 | www.nature.com/naturenanotechnology