Page 1
I.Mirebeau, S.Petit , A. Gukasov, J.Robert,thesis S.Guitteny, Laboratoire Léon Brillouin, CEA-Saclay
P.BonvilleDSM/IRAMIS/SPEC, CEA-Saclay
C.DecorseICMMO, Université Paris XI
H.Mutka, J.Ollivier, M.Boehm, P.SteffensInstitut Laue Langevin, Grenoble
A.Sazonov LLB, Aachen University
Magnetic structures and anisotropic excitationsin Tb2Ti2O7 spin liquid
Page 2
Tb2Ti2O7: a hot topic
Why is Tb2Ti2O7 (or TTO) so interesting ?
7 Posters at HFM’14Kermarrec Malkin Fennel Hallas KaoSazonovYin
Page 3
Tb2Ti2O7: a hot topic
because nobody fully understands it!
TTO
quantum spin iceSpin
liquid
Antiferro-magnetic spin ice
magneto-elastic liquid
Spin Glass
Page 4
Tb2Ti2O7: a hot topic
More and more sophisticated experiments
Influence of tiny defects
Coupling with the lattice
In the last 3 years
• Searching for a magnetization plateau : H //111• Probing dispersive excitations
• ½ ½ ½ structure• Competing SRO structures : Spin glass like vs. mesoscopic order
• magneto-elastic mode• Dynamic Jahn-Teller transition and/or interactions between quadrupolar moments
Towards a more realistic description ?
Page 5
Mc. Clarthy- Gingras Rev Modern Phys. ( Dipolar Spin ices: The Ising case
R2Ti2O7 pyrochlores R=Dy, Ho Effective interaction Jeff = J+Ddip > 0
Dipolar spin iceAF
FeF3
4in-4out
Spin ice Den Hertog et al Phys. Rev. Lett. (1999)
Bramwell et al Phys. Rev. Lett (2000)
Tb
DyHo
Tb nearby the thresholdQuantum fluctuations at play: « quantum spin ice » Molavian, Gingras, Canals, PRL (2008)Molavian , Clarthy, Gingras arxiv0912.2957Mc. Clarthy- Gingras Rev Progress Physics 77 056501(2014)
What about the Crystal field ?
Page 6
The crystal field
Δ = 200 – 300K Ho, Dy spin ices
Δ = 10-20K (Tb)
Tb3+ is a non-Kramers ion
Strong but finite <111> anisotropy
Δ ~ 1.5 meV
=
=
= -
• No exchange fluctuations allowed within the GS doublet
• No intensity scattered by neutrons
Gingras, PRB (2000)Bonville, IM, PRB( 2007) Bertin,Chapuis, JPCM(2012) Zhang, Fritsch, PRB (2014) Klekovina- Malkin J Opt. Phys. (2014)
Cao et al PRL(2009)
Page 7
Δ ~ 1.5 meV
= + = +
I α I α
h: molecular field
Splitting of the Ground state doublet
In molecular field approach
Δ ~ 1.5 meV
= =
- =
But =0
dh
Quantum mixing in the GS.
1st order perturbation 0th order perturbation
Simplest case: entangled wave functions
(gjµBh/)2 (0.75/15)22.10-3
D: quantum mixing
gJµB/kB= 1 for Tb !
Page 8
Δ ~ 1.5 meV
= + = +
h: molecular field
Splitting of the Ground state doublet
In molecular field approach
Δ ~ 1.5 meV
= =
dh
Quantum mixing in the GS.
1st order perturbation 0th order perturbation
Simplest case: entangled wave functions
Virtual crystal field model
• Very small intensity associated with GS fluctuations (with resp. to CF )
• Spin ice anisotropy: magnetization plateau
Two singlet ground state
• each singlet is non magnetic : no static signal• the transition has a large spectral weight • Jahn-Teller distortion?
Molavian, Gingras, Canals PRL(2007)Molavian, McClarthy, Gingras arxiv(2009)
Bonville et al PRB(2011), PRB (2014)
Page 9
Searching for a magnetization plateauUsing Magnetization, susceptibility, MuSR : a controversial situation
low field anomalies of the susceptibility:
MuSR Baker PRB (2012)
Legl et al PRL (2012)
No plateau in the isothermal magnetization
cross over regime in the dynamics
Yin et al PRL(2013)
Lhotel et al PRB-RC (2012)
Spin glass-like freezing ? TF~200-400 mKFritsch , PRB(2014)
Page 10
Searching for a magnetization plateauUsing neutrons : magnetic structure for H//111
• Exclude all-in all out structure
• Gradual reorientation of the Tb moments in the Kagome plane (keeping 1in- 3 out) without Kagome ice structure
See poster A. Sazonov
Page 11
Searching for a magnetization plateau
• No evidence for the 1/3 plateau at ~2µB expected at very small fields (down to 80mK)
• quantitative agreement with MF model assuming a dynamical JT distortion:
• 4 moment values and angles• M(H) for H//100, 111, 110
Field Irreversibilities
Spin glass like freezing?
A. Sazonov et al PRB(2013)
D=0 no mixing
• see poster A. Sazonov
Page 12
Spin fluctuations at very low temperatureUsing unpolarized neutrons
2 components in the neutron cross section• elastic (dominant) • inelastic (low energy)
elastic
• Pinch points• diffuse maxima at ½ ½ ½ positions
inelastic
• becomes structured at low T• well accounted for by 2 singlet model + anisotropic
exchange
D=0.25K
See also:
Takatsu et al. JPCM (2011)
Fritsch et al PRB(2013)
Static character not reproduced by the 2 singlet model
Page 13
diffuse scatteringb = -0.13T/µB ; DQ=0.25K
Phase diagram
P. Bonville et al Phys. Rev. B (2011)
3d-map Experiment
Simulation6T2 ( LLB)
The main features of the diffuse scattering are reproduced
Simulation with• anisotropic exchange• dipolar interactions• CF• JT distortion along equivalent 100,
010, 001 cubic axes.( preserves the overall cubic symmetry)
• Dynamical JT (average Structure factors and not intensities)
Energy integrated intensity
Page 14
- 50 mK - 50 mK
S.Petit & al, PRB 86 (2012) T.Fennell & al, Science 326 (2009)
Q dependence of the elastic scattering • Pinch points in both compounds: Coulomb phase
strong spectral weight at Q=0 no spectral weight at Q=0 ½ ½ ½ maxima : AF correlations
Page 15
Analysis of the pinch points Strongly anisotropic correlations of algebric nature
conservation law in TTO spin liquid analogous to the ice rules
What are the spin component involved?
S.Guitteny & al, PRL 111 (2013)
T. Fennell et al PRL(2012)
Page 16
Polarization analysisFennell Science (2009) : Ho2Ti2O7
PRL (2013) Tb2Ti2O7
Longitudinal polarimetry separates spin components
xZ //110
x// Q
1
2
3 4
1’
2’
Neutron cross section
• Correlations along Q (or x)• between spin components M┴Q
Ho2Ti2O7
NSF: correlations « up-down » 1-1’ or 2-2’: Weak (2 Spins, between T)
SF: correlations « 2in-2 out » 1-2-3-4: Strong (4 spins, in a T)
Q
z
yMz
My
neutron polarization P// Z
• Non spin flip: N+ <MZ.Mz>• Spin Flip <My.My>
Page 17
Polarization +energy analysisFennell Science (2009) : Ho2Ti2O7
PRL (2013) Tb2Ti2O7
Q
z
yMz
Myx
Z //110
x// Q
1
2
3 4
1’
2’
Tb2Ti2O7 Look at the dispersion
Mz: « up-down » correlations: relaxing (Quasi-E)My: « 2 in-2out » correlations : dispersing (Inel.)
T=50 mK
Longitudinal polarimetry separates spin components
Neutron cross section
• Correlations along Q (or x)• between spin components M┴Q
neutron polarization P// Z
• Non spin flip: N+ <MZ.Mz>• Spin Flip <My.My>
Page 18
18
Low energy excitations
• In all directions • Quasi-élastic• Strong fluctuations
My• Along (h,h,h)
• quasi-élastic• along (h,h,2-h) et (h,h,0)
• propagating excitation• no gap (Δres = 0,07meV)• Disperses up to 0,3 meV• intensity varies like 1/ω
First observation of a dispersive excitation in fluctuating disordered medium
Mz
S. Guitteny et al PRL(2013)
Page 19
Nature of the static SRO? the ½ ½ ½ order
½ ½ ½ diffuse maxima• Short range ~8-10 A• below ~0.4K• Vanish in a small field ( ~200G)
Fennel PRL (2012)Fristch PRB(2012)Petit PRB (2012)
In single crystals
In powders½ ½ ½ Mesoscopic structure• Over 30-50A• Associated with Cp anomaly• tuned by minute defects in Tb content
Taniguchi PRB RC(2011)
Short range vs. mesoscopic order
See also poster E. Kermarrec
Page 20
powder samples Tb2+xTi2-xO7+y
½ ½ ½ ½ ½ 3/2
½ ½ 5/2
3/2 3/2 1/2
X=0
Mesoscopic structure for x=0 and x=0.01
Difference pattern: I(50 mk)- I(1K)
T=50mK
N
X=0
exp: P. Dalmas de Réotier 2 q (deg)
Neu
tron
cou
nts
Page 21
Symmetry analysis 2 orbits with no common IR
site 1
Sites 2-4
N site1 0 0 0 2 ¾ ¼ ½ 3 ¼ ½ ¾4 ½ ¾ ¼
space group Fd-3M, K= ½ ½ ½
Champion, PRB (2001) Stewart, Wills JPCM(2004) Gd2Ti2O7
No way to build a strong ½ ½ ½ peak for Ising spins!
Needs to break either Ising anisotropy or cubic symmetry
K // local <111> axis no intensity at ½ ½ ½
• No vectors of the IR along the local <111> axes• Contributions to ½ ½ ½ cancel by symmetry
Systematic search of magnetic structures • 1T • cfc translations (cubic cell : a)• K= ½ ½ ½ (magnetic unit cell: 2a)
Page 22
The best structures (x=0)moments remain close to local <111>axes (3-10 deg)
M=1.9(4) µB/Tb; Lc =60 A (Y=1.4)
X=0X=0
Correlation length ~30 -50 A
« Monopole layered structure » « AF -Ordered spin ice »
Page 23
Ferrimagnetic piling of SI Tetrahedra
moments remain close to local <111>axes (<10 degs)
Fritsch PRB (2012)
The best structures (x=0)
« AF -Ordered spin ice » « Monopole layered structure »
AF packed OSI cubic cells,
MZ
Z//001
S. Guitteny (thesis) derived from Tb2Sn2O7 I. M et al PRL (2005)
Page 24
Ferrimagnetic piling of SI Tetrahedra separated by monopole layers
moments remain close to local <111>axes (<10 degs)
Fritsch PRB (2012)
The best structures (x=0)
« AF -Ordered spin ice » « Monopole layered structure »
AF packed OSI cubic cells, separated by SI tetrahedra with M
Full of monopoles, but compatible with a distortion No monopoles, but symmetry breaking at each cubic cell no possible LRO?
MZMZ
Z//001
Page 25
Calculated diffuse scatteringIn a single crystal, correlation length reduced to 2 cubic cells
h, h, 0
0, 0
, l
0, 0
, l
h, h, 0
« Monopole layered structure » « AF -Ordered spin ice »
Experiments
Petit PRB (2013)Fennel PRL (2013)Fritsch PRB(2013)
1
2
3
4
1 2 3 41 2 3 4
1
2
3
4
Page 26
The ½ ½ ½ order: summary• ½ ½ ½ order cannot propagate without breaking the cubic symmetry
• different structures and/or K orientations may compete (in space, time) yielding:
• SRO (single crystal) • mesoscopic orders (powders, tuned by x)• Spin glass like irreversibilities : Yin (2013), Fritsch PRB (2014) , Lhotel (2013)
• 2 physical mechanisms at play for the magnetic excitations• Relaxation (quasielastic)• Dispersive excitations
• Analog to the double dynamics in SP particles or quantum molecular magnets
Magneto-elastic modes as a switching mechanism?
Quasielastic or slow relaxations (thermally activated ,QT)
Inelastic modes
Page 27
Probing the magneto-elastic coupling Interaction between 1st excited CF doublet and acoustic phonon branch
Guitteny PRL(2013)
see also:Fennel PRL(2013)this conf. M. Ruminy : next talk
Other probes• pressure induced magnetic orderIM et al Nature 2002, PRL(2004)
•Elastic constantsKlekovina-Malkin J. Phys. 2011, J. Opt. Phys. 2014
•Thermal conductivityLi et al PRB(2013)
Page 28
Summary: what is new in TTO?• Quantum mixing in the GS doublet due to quadrupolar order: a necessary ingredient
• JT distortion « exchange » int. between quadrupolar moments
• Magnetoelastic coupling
• Non-Kramers character is crucial
• First observation of dispersive anisotropic excitations in a fluctuating disordered medium Two types of dynamics : relaxation, excitations
• Competing SI correlations with K=½ ½ ½• Not compatible with cubic symmetry• Tuned by off-stoechiometries• With different time and length scales• Associated with glassy behaviour
Gehring-Gehring (1985) Savary-Balents PRL(2012) Lee-Onoda-Balents PRB(2012)
MF
poster Malkin
Page 29
x=0.01
coexistence of LRO and mesoscopic orders
• Mesoscopic: M= 1.3µB/Tb
• LRO: M=0.3 µB/Tb
Page 30
I.M et al Nature (2002)
Under pressure : a phase with larger unit cell is also stabilized
Pressure induced structures