Вы находитесь на странице: 1из 5

RE S EAR CH | R E P O R T S

14. D. Best, H. W. Lam, J. Org. Chem. 79, 831–845 22. E. Marcantoni, M. Petrini, In Lewis Acid Promoted Addition Ministry of Education, Culture and Science Gravity program
(2014). Reactions of Organometallic Compounds (Elsevier, 2014), grant 024.001.035 (S.R.H.). We thank J. T. A. de Jong,
15. I. N. Houpis et al., Tetrahedron 54, 1185–1195 pp. 344–364. B. Maciá, and J. F. Collados for helpful comments on
(1998). 23. Y. Yamamoto, S. Yamamoto, H. Yatagai, Y. Ishihara, the manuscript.
16. G. Pattison, G. Piraux, H. W. Lam, J. Am. Chem. Soc. 132, K. Maruyama, J. Org. Chem. 47, 119–126 (1982).
14373–14375 (2010). 24. J. Rong, R. Oost, A. Desmarchelier, A. J. Minnaard,
17. A. Saxena, H. W. Lam, Chem. Sci. 2, 2326–2331 S. R. Harutyunyan, Angew. Chem. Int. Ed. 54, 3038–3042 SUPPLEMENTARY MATERIALS
(2011). (2015).
25. S. R. Harutyunyan et al., J. Am. Chem. Soc. 128, 9103–9118 www.sciencemag.org/content/352/6284/433/suppl/DC1
18. L. Rupnicki, A. Saxena, H. W. Lam, J. Am. Chem. Soc. 131,
(2006). Materials and Methods
10386–10387 (2009).
26. N. Yoshikai, E. Nakamura, Chem. Rev. 112, 2339–2372 Table S1
19. A. A. Friedman, J. Panteleev, J. Tsoung, V. Huynh, M. Lautens,
(2012). NMR spectra
Angew. Chem. Int. Ed. 52, 9755–9758 (2013).
20. H. G. Richey, Grignard Reagents: New Developments (Wiley, References (27–35)
1999). ACKN OWLED GMEN TS
21. H. Yamamoto, Ed., Lewis Acids in Organic Synthesis, Supported by Netherlands Organization for Scientific 6 January 2016; accepted 22 March 2016
Volume 1–2 (Wiley-VCH, 2000). Research NWO-Vidi and ECHO grants (S.R.H.) and 10.1126/science.aaf1983

MAGNETISM ner has turned out to be challenging. In this


work, we relied on the simplest (albeit demand-

Atomic-scale control of graphene ing) experimental approach to remove a single pz

Downloaded from http://science.sciencemag.org/ on May 31, 2016


orbital from the graphene network by means of
the adsorption of a single H atom. Atomic H
magnetism by using hydrogen atoms chemisorbs on graphene on top of carbon atoms,
changing the initial sp2 hybridization of carbon
to essentially sp3 (17, 18) and effectively removing
Héctor González-Herrero,1,2 José M. Gómez-Rodríguez,1,2,3 Pierre Mallet,4,5 the corresponding pz orbital (4, 19, 20). In this
Mohamed Moaied,1,6 Juan José Palacios,1,2,3 Carlos Salgado,1 Miguel M. Ugeda,7,8 sense, chemisorbed H atoms are equivalent to
Jean-Yves Veuillen,4,5 Félix Yndurain,1,2,3 Iván Brihuega1,2,3* carbon vacancies (4, 12, 14) but with the advantage
that, unlike vacancies, they leave the graphene
Isolated hydrogen atoms absorbed on graphene are predicted to induce magnetic atomic lattice with no unsaturated dangling bonds,
moments. Here we demonstrate that the adsorption of a single hydrogen atom on preserving the threefold symmetry. Our exper-
graphene induces a magnetic moment characterized by a ~20–millielectron volt spin-split iments, supported by ab initio calculations, provide
state at the Fermi energy. Our scanning tunneling microscopy (STM) experiments, a comprehensive picture of the origin, coupling,
complemented by first-principles calculations, show that such a spin-polarized state is and manipulation of the magnetism induced by
essentially localized on the carbon sublattice opposite to the one where the hydrogen atom H atoms on graphene layers.
is chemisorbed. This atomically modulated spin texture, which extends several nanometers We deposited atomic H on graphene grown
away from the hydrogen atom, drives the direct coupling between the magnetic moments at on a SiC(000-1) substrate (21). In this system,
unusually long distances. By using the STM tip to manipulate hydrogen atoms with atomic the rotational disorder of the graphene layers
precision, it is possible to tailor the magnetism of selected graphene regions. electronically decouples the p bands, leading to

A
a stacking of essentially isolated graphene sheets
dding magnetism to the long list of graphene’s on the spatial localization of the state, because (22–24). Scanning tunneling microscopy (STM)
capabilities has been pursued since this ma- this defines the proximity of the electrons (Fig. visualizes single H atoms as a bright protrusion
terial was first isolated (1). From a theoret- 1A). In contrast to magnetic moments of a strong- (apparent height, ~2.5 Å) surrounded by a com-
ical point of view, magnetic moments in ly localized atomic character that are commonly plex threefold √3×√3 pattern that is rotated 30°
graphene can be induced by removing a found in magnetic materials, these induced mo- degrees with respect to the graphene lattice. (25, 26)
single pz orbital from the p-graphene system; this ments are predicted to extend over several na- [Fig. 1B and section 1 in (27)]. The resolution that
removal creates a single p-state at the Fermi en- nometers, suggesting a strong direct coupling we achieved allowed us to identify the adsorbate as
ergy (EF) around the missing orbital. The double between them at unusually long distances. The a single H atom and to determine the atomic site
occupation of this state by two electrons with coupling rules between the induced magnetic (and thus the corresponding atomic sublattice)
different spins is forbidden by the electrostatic moments are also expected to be simple. Because where each H atom was chemisorbed by means of
Coulomb repulsion; namely, once an electron oc- of the bipartite atomic structure of graphene— comparison with density functional theory (DFT)–
cupies the state, a second one with opposite spin which consists of two equivalent triangular sub- simulated STM images [Fig. 1D and section 1 in
needs to “pay” an extra energy U. This leaves a lattices, labeled A and B—and according to Lieb’s (27)]. As depicted in Fig. 1A, graphene magnetic
single electron occupying the state and therefore theorem (7), the ground state of the system pos- moments induced by H adsorption should be
a net magnetic moment (2–6). The strength of U, sesses a total spin given by S = 1/2 × |NA–NB|, where reflected in the appearance of a spin-polarized
which determines the spin splitting, depends NA and NB are the number of pz orbitals removed state at EF, which, according to DFT calculations,
from each sublattice (4, 8, 9). Thus, to generate a should be characterized by two narrow peaks in
1
Departamento de Física de la Materia Condensada,
net magnetic moment in a particular graphene the density of states (DOS) (Fig. 1E) (4).
Universidad Autónoma de Madrid, E-28049 Madrid, Spain. region, a different number of pz orbitals from Differential conductance spectra (dI/dV; I, cur-
2
Condensed Matter Physics Center (IFIMAC), Universidad each sublattice needs to be locally removed. rent; V, sample voltage) probe the energy-resolved
Autónoma de Madrid, E-28049 Madrid, Spain. 3Instituto Many theoretical proposals have been put for- local DOS under the STM tip position and
Nicolás Cabrera, Universidad Autónoma de Madrid, E-28049
Madrid, Spain. 4Université Grenoble Alpes, Institut NEEL, F-
ward on this subject, involving zigzag edges, thus are ideal for investigating this question.
38042 Grenoble, France. 5Centre National de la Recherche graphene clusters, grain boundaries, and atomic Figure 1C shows two dI/dV spectra, measured
Scientifique (CNRS), Institut NEEL, F-38042 Grenoble, defects (2, 4, 5, 8–11). Experimentally, the remov- at 5 K, that are representative of our findings.
France. 6Department of Physics, Faculty of Science, Zagazig al of pz orbitals from the p system has been The dI/dV spectra measured on clean graphene,
University, 44519 Zagazig, Egypt. 7CIC nanoGUNE, 20018
Donostia-San Sebastian, Spain. 8Ikerbasque, Basque
achieved by randomly creating atomic vacan- located far enough away from defects, have the
Foundation for Science, 48013 Bilbao, Spain. cies or adsorbing adatoms (12–16). However, characteristic featureless V shape of graphene,
*Corresponding author. Email: ivan.brihuega@uam.es removing those pz orbitals in a controlled man- with a minimum at EF indicating the position of

SCIENCE sciencemag.org 22 APRIL 2016 • VOL 352 ISSUE 6284 437


R ES E A RC H | R E PO R TS

the Dirac point ED. The dI/dV spectra measured which is in agreement with the experimental n- and p-type graphene doping cause the split-
on top of single H atoms have two narrow peaks, observations. ting of the H-induced graphene state to vanish;
one below and one above EF, separated in energy An independent proof that the split dI/dV only one sharp peak appears at ED, which we
by a splitting of ~20 meV. We attribute these two peaks are induced by magnetism can be obtained ascribe to a transition to the nonmagnetic state.
features to the spin-polarized state, in which the by changing the occupation of the split states (n↓, This interpretation is fully supported by our DFT
Coulomb repulsion is large enough to fully sep- n↑) by means of graphene doping. This is based calculations for doped graphene layers [Fig. 1F
arate the two spin components. The observed on the idea originally proposed in (28), according and section 3B in (27)]. If the split peaks appear-
charge neutrality (the splitting is essentially sym- to which the transition from a magnetic state to ing in neutral graphene after H adsorption had
metric around EF) and the well-defined peak a nonmagnetic one can be realized by tuning the an origin that was not associated with a magnetic
splitting indicate the complete spin polarization energy position of the impurity level with respect moment (i.e., a single-electron origin), we would
of the state. DFT calculations show that the to the Fermi level [section 3A in (27) gives a de- observe a rigid shift of the peak position with
magnetic moment associated with the unpaired tailed description of this system in terms of the doping, and the doublet structure would remain
electron that is left over in the graphene system Anderson impurity model]. In graphene, the im- unmodified [section 3 in (27)]. Our results are con-
after H adsorption would be 1 bohr magneton purity level (zero-energy mode) should be dic- sistent with the case of sp3 defects in graphene,
(fig. S7). Our interpretation of the experiment is tated by the position of ED [E↑ = ED + U(n↓ – 1/2); for which the possibility to control graphene
fully supported by DFT calculations, as can be E↓ = ED + U(n↑ – 1/2)] (29), which can be tuned magnetic moments by molecular doping has been
seen in Fig. 1E, which shows the expected DOS by doping the graphene layers. For a large enough reported (15).
for a single H atom in a graphene super cell electron (hole) doping, the occupation of both We explored, with atomic precision, the spatial
containing 218 carbon atoms. The theoretical the n↑ and n↓ levels can be tuned to 1 (0), in which extension of the spin-polarized electronic state

Downloaded from http://science.sciencemag.org/ on May 31, 2016


energy splitting depends on the size of the gra- case the energy levels will be degenerate, leading induced by H atoms in undoped graphene. The
phene super cell (5). Our calculations show that to a single nonmagnetic state close to ED (E↑ = relatively modest 20-meV energy splitting ob-
the splitting decays with the size of the gra- E↓ ≈ ED). In Fig. 1F, we show how graphene doping served in our experiments suggests a large spa-
phene super cell, suggesting a small but finite affects the splitting of the H-induced magnetic tial extension of the magnetic state (Fig. 1A).
splitting for the isolated H [section 9 in (27)], state. Our dI/dV spectra demonstrate that both Figure 2A shows a conductance map plotted with

Fig. 1. Spin-split state induced by atomic H on graphene. (A) Illustra- were acquired at a nominal junction impedance of 2 gigaohms (–100 mV,
tion of the origin of the spin-split state in terms of its spatial extension (r) 50 pA). (D and E) DFT-simulated STM image (D) and DOS (E) of an H
given by the square of its wave function y2, and the corresponding elec- atom chemisorbed on neutral graphene. (F) dI/dV spectra and DFT cal-
trostatic Coulomb repulsion U. Arrows indicate the energy position of culation of the DOS induced by a single H atom on n- (left) and p-type
spin-up (n↑) and spin-down (n↓) levels. For a fully polarized one-electron (right) doped graphene. The minimum of the dI/dV spectra, acquired on
state, the majority level spin is filled and the minority one is empty; there- bare graphene (black curves), determines the position of ED. The spectra
fore, n↑ = 1 and n↓ = 0, and the energy splitting (E↓ – E↑) is given by U were acquired at a nominal junction impedance of 8 gigaohms (–400 mV,
[section 3 in (27)]. (B) STM topography of a single H atom chemisorbed 50 pA). STM data were acquired and processed using the WSxM software
on neutral graphene (0.2 V, 0.1 nA, 7 × 7 nm2). (C) dI/dV spectrum mea- (36). Calculations for all simulated images were performed at the same
sured on the H atom, showing a fully polarized peak at EF, and measured energy as in the corresponding experimental conditions. All experimental
on bare graphene far from the H atom (a.u., arbitrary units). The spectra data were acquired at 5 K.

438 22 APRIL 2016 • VOL 352 ISSUE 6284 sciencemag.org SCIENCE


RE S EAR CH | R E P O R T S

respect to distance and energy [dI/dV(x, E)], with in the same sublattice as H chemisorption, the atoms in the graphene sublattice opposite to the
the distance x measured along the 6-nm line that peaks vanish almost completely (fig. S11). Because locus of H chemisorption.
crosses the H atom in Fig. 2B. The state extends our results show a complete spin polarization of The large extension of the local magnetic mo-
several nanometers away from the H atom, con- the state, the spatial evolution of the height of the ments associated with H chemisorption suggests
firming that it is a quasi-localized graphene state dI/dV occupied peak provides the spatial distri- that long-range magnetic interactions mediated
(3). It presents strong atomic-scale modulations of bution of the local magnetic moment induced by by direct exchange should take place. This is dif-
the peak intensities, with maxima (bright yellow H chemisorption [DFT calculations in Fig. 2, C ferent from substrate-mediated interactions such
features in the conductance map) corresponding and D, and section 4 in (27)]. This is further il- as the Ruderman-Kittel-Kasuya-Yosida interac-
to carbon atoms in the sublattice opposite to the lustrated in Fig. 2E, which shows that the mag- tion, because in this case the coupling results
one on which H is chemisorbed. On carbon sites netic moment is essentially induced on the carbon from the direct overlap of magnetized graphene

Downloaded from http://science.sciencemag.org/ on May 31, 2016


Fig. 2. Spatial extension of the spin-polarized electronic state induced by moments induced by H chemisorption (the lengths of the arrows signify the relative
H atoms in undoped graphene. (A) Conductance map [dI/dV(x,E)] along the magnitudes of the magnetic moments). (E) Schematic of the graphene structure
dashed line in (B). The spectra were acquired at a nominal junction impedance of along the dashed line in (B). Green (purple) balls indicate the positions of carbon
3 gigaohms (100 mV, 33 pA). (B) STM topography of a single H atom on graphene atoms belonging to the same (opposite) sublattice with respect to the locus of H
(0.2 V, 0.1 nA, 7 × 5 nm2). (C) Comparison between DFT calculations for the local chemisorption. The dotted line shows the evolution of the height of the measured
magnetic moment and the height of the occupied projected DOS (PDOS) peak, occupied peak, and the arrows show the relative magnetic moment contribution of
calculated on different carbon atoms [section 4 in (27)]. (D) Calculated magnetic each carbon atom. All experimental data were acquired at temperature T = 5 K.

Fig. 3. Sublattice dependence of the magnetic coupling between neigh- configuration (0.2 V, 0.1 nA, 7.8 × 6.6 nm2). (C) Calculated STM image of the AB
boring H atoms. (A) Calculated total energy, relative to twice the adsorption dimer and (D) the AA dimer, with the corresponding diagrams for H atoms
energy of a single H atom, and magnetic state of a pair of H atoms adsorbed on (blue balls) on graphene (purple and green balls). (E) Experimental dI/dV spec-
the same (AA dimer) and different (AB dimer) sublattices, plotted as a function tra and (F) calculated DOS for the AA dimer, AB dimer, and clean graphene.
of the H-H distance. (B) STM image showing two different pairs of H atoms, The spectra were acquired at a nominal junction impedance of 8 gigaohms
with one pair in an AA (purple circle) and the other pair in an AB (green circle) (–400 mV, 50 pA). All experimental data were acquired at 5 K.

SCIENCE sciencemag.org 22 APRIL 2016 • VOL 352 ISSUE 6284 439


R ES E A RC H | R E PO R TS

Fig. 4. Manipulation of graphene local magnetic moments by STM. (A) STM dimer in (F). The spectra were acquired at a nominal junction impedance of
image of an H dimer in an AB configuration. (B) STM image after the removal of 4 gigaohms (200 mV, 50 pA). An additional dI/dV spectrum that is better

Downloaded from http://science.sciencemag.org/ on May 31, 2016


one H atom. (C) dI/dVspectra measured on the AB dimer in (A) and (D) the single resolved in the vicinity of EF measured on the AA dimer in (E) is shown in fig. S20.
H atom in (B). The spectra were acquired at a nominal junction impedance of Insets show the corresponding DFT calculations. (I to L) STM images showing
4 gigaohms (200 mV, 50 pA). The insets present the corresponding DFT the same graphene region during different steps of a manipulation experiment
calculations for H atoms (blue balls) on graphene (purple and green balls), involving a large number of H atoms.The point defect outlined with a gray circle is
with blue arrows being the magnetic moments induced on graphene. (E) STM used as a reference.Tunneling parameters were 0.2 V, 0.1 nA, and 6.5 × 4.0 nm2
image of an H dimer in an AA configuration. (F) STM image after laterally moving for (A) and (B); 0.2 V, 0.1 nA, and 9.5 × 5.5 nm2 for (E) and (F); and 0.4 V, 0.03 nA,
one H atom. (G) dI/dV spectra measured on the AA dimer in (E) and (H) the AB and 28 × 28 nm2 for (I) to (L). All experimental data were acquired at T = 5 K.

states. In addition, the critical sublattice depen- sured in this study, AA dimers presented a fully moments induced by this H dimer, we turned
dence that we observed for the spin-polarized split state close to EF, which was absent in AB it into a nonmagnetic AB dimer configuration
peak implies that the magnetic coupling should dimers (fig. S12). Our STS data show that this by laterally moving one of its H atoms to the
be radically different depending on whether H sublattice-dependent magnetic coupling persists opposite sublattice. Figure 4F shows the same
atoms are adsorbed on the same or different for very long distances, even for H dimers sep- graphene region after the H manipulation (the
sublattices. Consistent with this expectation, arated by more than 1 nm [section 5 in (27)]. AB dimer in the upper part of the image serves
our DFT calculations reveal that two H atoms We further demonstrated the capability of in- as reference). The STS spectrum measured
chemisorbed on the same sublattice (AA dimer) ducing magnetic moments on selected graphene on the constructed AB dimer shows the disap-
show ferromagnetic coupling with a total spin regions by using STM to perform atomic manip- pearance of the polarized peaks, indicating
S = 1, whereas for H atoms on different sub- ulations (33–35). We proved that individual H that local graphene magnetism was effectively
lattices (AB dimer), the solution is nonmagnetic. atoms can be removed, laterally moved, and even switched off.
This result is reproduced for all possible H-H ar- deposited on graphene surfaces with atomic pre- Last, we explored the possibility of selec-
rangements up to the largest distances (~1.5 nm) cision to ultimately tailor their local magnetic tively tuning the collective magnetic moment
achievable with our super cell sizes (Fig. 3A). For state [section 6 in (27)]. Figure 4 shows two rep- in a graphene region by inducing an imbalance
a separation of 1.5 nm, the computed exchange resentative examples of these manipulation exper- between H atoms on opposing sublattices A
energy for AA dimers is ~35 meV [section 5 in iments, in which the local graphene magnetism and B. For this purpose, we systematically ma-
(27)]. Furthermore, the total energy of all H dimers was selectively switched on and off. The graphene nipulated a large number of H atoms [section 6
that we studied is lower than that of two isolated H region in Fig. 4A shows two H atoms in an AB in (27)]. In Fig. 4, I to L, we present an example
atoms (Fig. 3A), confirming the observed tendency dimer configuration. Our STS data measured on in which we first removed all H atoms from a
of H to form dimers on graphene surfaces at high those H atoms (Fig. 4C) show that this AB dimer graphene region by using the STM tip (Fig. 4I).
enough H concentrations (30–32). configuration does not induce any magnetism Then, we selectively deposited 14 H atoms on
To test this scenario experimentally, we ex- on the graphene layer, in good agreement with this same region to reach a configuration with
plored the local electronic structure of many dif- the coupling rules previously discussed. Figure seven H atoms chemisorbed on each graphene
ferent H dimers with high-resolution scanning 4B shows the same graphene region as in Fig. sublattice (Fig. 4J). Our experimental findings
tunneling spectroscopy (STS). The STM image in 4A, after the controlled extraction of one H and existing calculations (4, 7) indicate that a
Fig. 3B shows two H dimers in AA (purple circle) atom by carefully approaching it with the STM very low (if any) net magnetic moment should
and AB (green circle) configurations; the corre- tip. As shown in Fig. 4D, a spin-split state im- be expected on this region, because of these
sponding calculated STM images are shown in mediately emerges on the graphene layer after equal sublattice populations. Next, by selectively
Fig. 3, C and D. The dI/dV spectra acquired on the H removal, confirming the creation of a local removing all the H atoms chemisorbed on sub-
the AB dimer (green line in Fig. 3E) show a magnetic moment in graphene. The insets show lattice B, we created a ferromagnetic configuration
featureless local DOS that is indistinguishable the corresponding DFT calculations of the result- with the seven remaining H atoms on sublattice
from that measured on bare graphene (black), as ing magnetic moment for each situation. We A (Fig. 4K). As the final step, we combined sev-
would be expected for a nonmagnetic config- next performed a lateral manipulation on the eral manipulation processes to reverse the sit-
uration. In contrast, dI/dV spectra measured on H dimer, shown in the central region of Fig. 4E. uation and construct an H arrangement with all
the AA dimer (purple) show the split state in the Initially, the dimer was in an AA configuration, seven H atoms chemisorbed on sublattice B (Fig.
vicinity of EF, as expected for a ferromagnetic with both H atoms chemisorbed on the same 4L). The degree of complexity shown in our ma-
coupling between the H atoms. As shown in Fig. 3F, carbon sublattice. The STS spectrum for that nipulation experiments demonstrates the high
our calculated DOS reproduce these observations, configuration (Fig. 4G) shows the presence of a reproducibility of the procedure, which paves the
confirming the ferromagnetic (nonmagnetic) na- spin-split state, as expected for ferromagnetic way to the realization of atomically controlled
ture of the AA (AB) dimer. For all H dimers mea- coupling. To switch off the graphene magnetic experiments in graphene magnetism, an area

440 22 APRIL 2016 • VOL 352 ISSUE 6284 sciencemag.org SCIENCE


RE S EAR CH | R E P O R T S

that has thus far been restricted to a purely theo- QUANTUM OPTICS
retical framework.
RE FE RENCES AND N OT ES
1. K. S. Novoselov et al., Science 306, 666–669 (2004).
Bell correlations in a
2. P. O. Lehtinen, A. S. Foster, Y. Ma, A. V. Krasheninnikov,
R. M. Nieminen, Phys. Rev. Lett. 93, 187202 (2004).
3. V. M. Pereira, F. Guinea, J. M. Lopes dos Santos, N. M. R. Peres,
Bose-Einstein condensate
A. H. Castro Neto, Phys. Rev. Lett. 96, 036801 (2006).
4. O. V. Yazyev, L. Helm, Phys. Rev. B 75, 125408 (2007). Roman Schmied,1* Jean-Daniel Bancal,2,4* Baptiste Allard,1* Matteo Fadel,1
5. O. V. Yazyev, Rep. Prog. Phys. 73, 056501 (2010).
6. E. J. G. Santos, A. Ayuela, D. Sanchez-Portal, New J. Phys. 14, Valerio Scarani,2,3 Philipp Treutlein,1† Nicolas Sangouard4†
043022 (2012).
7. E. H. Lieb, Phys. Rev. Lett. 62, 1201–1204 (1989). Characterizing many-body systems through the quantum correlations between their
8. J. Fernández-Rossier, J. J. Palacios, Phys. Rev. Lett. 99, 177204 (2007).
constituent particles is a major goal of quantum physics. Although entanglement is routinely
9. J. J. Palacios, J. Fernández-Rossier, L. Brey, Phys. Rev. B 77,
195428 (2008). observed in many systems, we report here the detection of stronger correlations—Bell
10. M. Fujita, K. Wakabayashi, K. Nakada, K. Kusakabe, J. Phys. correlations—between the spins of about 480 atoms in a Bose-Einstein condensate. We
Soc. Jpn. 65, 1920–1923 (1996). derive a Bell correlation witness from a many-particle Bell inequality involving only one- and
11. C. G. Tao et al., Nat. Phys. 7, 616–620 (2011).
two-body correlation functions. Our measurement on a spin-squeezed state exceeds the
12. M. M. Ugeda, I. Brihuega, F. Guinea, J. M. Gómez-Rodríguez,
Phys. Rev. Lett. 104, 096804 (2010). threshold for Bell correlations by 3.8 standard deviations. Our work shows that the strongest
13. R. R. Nair et al., Nat. Phys. 8, 199–202 (2012). possible nonclassical correlations are experimentally accessible in many-body systems and

Downloaded from http://science.sciencemag.org/ on May 31, 2016


14. K. M. McCreary, A. G. Swartz, W. Han, J. Fabian, that they can be revealed by collective measurements.
R. K. Kawakami, Phys. Rev. Lett. 109, 186604 (2012).

P
15. R. R. Nair et al., Nat. Commun. 4, 2010 (2013).
16. J. Balakrishnan, G. Kok Wai Koon, M. Jaiswal, A. H. Castro Neto, arts of a composite quantum system can Whereas all Bell-correlated states are entangled,
B. Özyilmaz, Nat. Phys. 9, 284–287 (2013). share correlations that are stronger than the converse is not true (1). In recent theoretical
17. L. Jeloaica, V. Sidis, Chem. Phys. Lett. 300, 157–162 (1999). any classical theory allows (1). These so- work, a family of Bell inequalities was derived
18. X. W. Sha, B. Jackson, Surf. Sci. 496, 318–330 (2002).
19. D. W. Boukhvalov, M. I. Katsnelson, A. I. Lichtenstein, Phys.
called Bell correlations represent the most that are symmetric under particle exchange and
Rev. B 77, 035427 (2008). profound departure of quantum from clas- involve only first- and second-order correlation
20. D. Soriano et al., Phys. Rev. Lett. 107, 016602 (2011). sical physics and can be confirmed experimen- functions (23). It was suggested that this could
21. F. Varchon, P. Mallet, L. Magaud, J. Y. Veuillen, Phys. Rev. B 77, tally by showing that a Bell inequality is violated enable the detection of Bell correlations by col-
165415 (2008).
22. M. Sprinkle et al., Phys. Rev. Lett. 103, 226803 (2009).
by the system. The existence of Bell correlations lective measurements on spin ensembles. Acting
23. G. H. Li et al., Nat. Phys. 6, 109–113 (2010). at spacelike separations refutes local causality on this proposal, we derive a collective witness
24. G. Trambly de Laissardière, D. Mayou, L. Magaud, Nano Lett. (2); thus, Bell correlations are also called non- observable that is tailored to detect Bell correla-
10, 804–808 (2010). local correlations. Moreover, they are a key re- tions in spin-squeezed states of atomic ensem-
25. H. A. Mizes, J. S. Foster, Science 244, 559–562 (1989).
26. P. Ruffieux, O. Groning, P. Schwaller, L. Schlapbach,
source for quantum technologies such as quantum bles. We report a measurement of this witness
P. Groning, Phys. Rev. Lett. 84, 4910–4913 (2000). key distribution and certified randomness gen- on 480 ultracold rubidium atoms, revealing Bell
27. Materials and methods are available as supplementary eration (3). Bell correlations have so far been correlations in a many-body system.
materials on Science Online. detected between up to 14 ions (4), four photons We derive our Bell correlation witness in
28. P. W. Anderson, Phys. Rev. 124, 41–53 (1961).
29. B. Uchoa, V. N. Kotov, N. M. R. Peres, A. H. Castro Neto, Phys.
(5, 6), two neutral atoms (7), two solid-state spin the context of a Bell test where N observers (in-
Rev. Lett. 101, 026805 (2008). qubits (8), and two Josephson phase qubits (9). dexed by i ¼ 1…N ) each repeatedly perform one
ðiÞ ðiÞ
30. L. Hornekær et al., Phys. Rev. Lett. 97, 186102 (2006). Even though multipartite Bell inequalities are of two possible local measurements M0 or M1
31. Z. Sljivančanin et al., J. Chem. Phys. 131, 084706 (2009). known (1, 10–12), the detection of Bell correla- on their part of a composite system and observe
32. R. Balog et al., J. Am. Chem. Soc. 131, 8744–8745 (2009).
33. D. M. Eigler, E. K. Schweizer, Nature 344, 524–526 (1990). tions in larger systems is challenging. one of two possible outcomes ai ¼ T1. For ex-
34. T. C. Shen et al., Science 268, 1590–1592 (1995). A central challenge in quantum many-body ample, the system could be an ensemble of
35. A. A. Khajetoorians et al., Science 339, 55–59 (2013). physics is to connect the global properties of a atomic spins where each observer is associated
36. I. Horcas et al., Rev. Sci. Instrum. 78, 013705 (2007).
system to the underlying quantum correlations with one atom and the measurements corre-
ACKN OW LEDG MEN TS between the constituent particles (13, 14). For spond to spin projections along different axes.
We thank V. Cherkez (Institut NEEL, CNRS, and Université example, recent experiments in quantum me- When all observers choose to measure M0 , one
Grenoble Alpes) for his help with the fabrication of samples trology have shown that spin-squeezed states determines experimentally
P the sum of their aver-
ðiÞ
of graphene grown on SiC, and we thank D. Wong (University of of atomic ensembles can enhance the precision age outcomes S 0 ¼ N i¼1 hM0 〉 and correlations
California–Berkeley) for his careful reading of the manuscript. This
of interferometric measurements beyond clas- PN ðiÞ ðjÞ
work was supported by Spain’s Ministerio de Economía y
sical limits (15–18). This enhancement requires S 00 ¼ i;j¼1ði≠jÞ hM0 M0 i [see section 1 of the
Competitividad under grant nos. MAT2013-41636-P, CSD2010-
00024, PCIN-2015-030, FIS2013-47328, and FIS2012-37549-C05- entanglement between atoms in the ensemble, supplementary materials (24) for a definition
03; the European Union structural funds and the Comunidad de which can be revealed by measuring an entan- in terms
Madrid MAD2D-CM program under grant nos. S2013/MIT-3007 P of measured ðiÞ
frequencies]. Similarly,
ðjÞ
and P2013/MIT-2850; the Generalitat Valenciana under grant
glement witness that involves only collective S 11 ¼ N i; j¼1ði≠jÞ hM1 M1 i is determined when
no. PROMETEO/2012/011; THE CNRS PICS (Projets Internationaux measurements on the entire system (15, 19–22). all observers choosePN M1 . A more
ðiÞ
complex cor-
ðjÞ
de Coopération Scientifique) program under grant no. 6182; and The role of Bell correlations in many-body sys- relation S 01 ¼ i; j¼1ði≠jÞ hM0 M1 i is quan-
the European Union FP7 (7th Framework Programme for Research tems, on the other hand, is largely unknown. tified by letting all pairs of observers choose
and Technological Development) Graphene Flagship (grant
604391) and FLAG-ERA programs. The authors acknowledge the
opposite measurements, which requires repeated
computer resources and assistance provided by the Centro de observations of identically prepared states of the
1
Computación Científica of the Universidad Autónoma de Madrid. Quantum Atom Optics Laboratory, Department of Physics, system because some of these measurements are
University of Basel, Klingelbergstrasse 82, 4056 Basel, mutually exclusive. In (23), a Bell inequality was
SUPPLEMENTARY MATERIALS Switzerland. 2Centre for Quantum Technologies, National
University of Singapore, 3 Science Drive 2, Singapore
derived that contains only these symmetric one-
www.sciencemag.org/content/352/6284/437/suppl/DC1
Materials and Methods 117543. 3Department of Physics, National University of and two-body correlators.
Figs. S1 to S20 Singapore, 2 Science Drive 3, Singapore 117542. 4Quantum
1 1
Movie S1 Optics Theory Group, Department of Physics, University of 2S 0 þ S 00 þ S 01 þ S 11 þ 2N≥ 0 ð1Þ
References (37–59) Basel, Klingelbergstrasse 82, 4056 Basel, Switzerland. 2 2
*These authors contributed equally to this work. †Corresponding
3 November 2015; accepted 21 March 2016 author. Email: philipp.treutlein@unibas.ch (P.T.), nicolas.sangouard@ If an experiment violates this inequality, the
10.1126/science.aad8038 unibas.ch (N.S.) conditional probabilities Pða1 ; …; aN jx1 ; …; xN Þ

SCIENCE sciencemag.org 22 APRIL 2016 • VOL 352 ISSUE 6284 441

Вам также может понравиться