22.01.2015 Views

Knight Shift in Metals

Knight Shift in Metals

Knight Shift in Metals

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

(1)<br />

Nuclear Magnetic Resonance (NMR)<br />

Part 2: <strong>Knight</strong> <strong>Shift</strong> <strong>in</strong> <strong>Metals</strong><br />

shift of MNR resonance l<strong>in</strong>e due to polarisation of conduction electrons.<br />

Consider polarisation of conduction electrons <strong>in</strong> a magnetic field B 0 .<br />

B0 0<br />

E<br />

sp<strong>in</strong> antiparallel B <br />

sp<strong>in</strong> parallel B <br />

EF<br />

}<br />

2 B B0<br />

E<br />

F<br />

k<br />

1 meV<br />

110eV<br />

<br />

excess of electrons with s B<br />

<br />

0<br />

<br />

M n n<br />

net magnetisation <br />

<br />

0<br />

<br />

<br />

e<br />

B<br />

<br />

1<br />

n<br />

f E DE B<br />

B0<br />

dE<br />

2<br />

<br />

f E<br />

electron density parallel and antiparallel to B0<br />

1<br />

E<br />

EF<br />

<br />

1exp <br />

"Fermi distribution function"<br />

kT B <br />

<br />

f E<br />

<br />

k<br />

B<br />

T<br />

E<br />

<br />

D E B D E D E<br />

B<br />

0<br />

B<br />

E F


(2)<br />

e<br />

<br />

2<br />

B <br />

0<br />

2<br />

B<br />

<br />

M f E D E dE<br />

<strong>in</strong>tegration by parts<br />

<br />

0<br />

<br />

<br />

<br />

B f<br />

E D E dE<br />

<br />

0<br />

DE <br />

F<br />

2<br />

3N<br />

Me B B0<br />

DEF DEF<br />

2E<br />

T <strong>in</strong>dependent s<strong>in</strong>ce E k T<br />

F<br />

0<br />

F<br />

<br />

pauli<br />

Me<br />

3N<br />

<br />

<br />

B<br />

2E<br />

<br />

2<br />

B<br />

2<br />

B<br />

F<br />

<br />

D E<br />

F<br />

<br />

Pauli sp<strong>in</strong> susceptibility of conduction electrons<br />

T-<strong>in</strong>dependent <strong>Knight</strong> shift<br />

<br />

e at the<br />

nuclear site!<br />

this is not necessarily the average one.<br />

15<br />

Nuclei are very small a 10 m . They are surrounded by a cloud of polarized<br />

conduction electrons.<br />

however, the <strong>Knight</strong> shift is determ<strong>in</strong>ed by magnetic field B r 0<br />

Need to sum up the contributions to the local magnetic field at the nucleus site.<br />

Consider homogenously magnetised sphere of magnetisation M<br />

2<br />

out 3<br />

e r<br />

r<br />

B r<br />

r a<br />

5<br />

r<br />

<br />

<strong>in</strong> 2<br />

B r<br />

<br />

r a<br />

3<br />

a<br />

have to consider lima<br />

0<br />

dipolar terms<br />

"contact-term"


(3)<br />

<br />

volume<br />

<br />

<strong>in</strong> 3 2<br />

4 3 8<br />

<br />

B d r const a<br />

<br />

3<br />

a 3 3<br />

so-called contact term:<br />

<br />

8<br />

3<br />

<br />

r<br />

<br />

<br />

2<br />

for polarized electrons: er<br />

0 el 0<br />

e Sz<br />

<br />

density of electrons at the nucleus site<br />

Hhf N Be<br />

2<br />

3erNrNe<br />

r<br />

<br />

5<br />

r<br />

8<br />

2 <br />

el 0 N re<br />

r<br />

r<br />

3 <br />

<br />

<br />

S I<br />

N e el<br />

<br />

<br />

Paschen-Back regime<br />

N e Sz Iz<br />

Me<br />

with e Sz e<br />

<br />

N<br />

pauli<br />

B0<br />

N<br />

N S D E<br />

e<br />

2<br />

pauli z B F<br />

B0<br />

<br />

<br />

N = electron density<br />

<strong>Knight</strong> <strong>Shift</strong> K<br />

0<br />

0<br />

Be<br />

8 el<br />

1<br />

K B 3 N <br />

pauli<br />

Important: measure susceptibility at nuclear site.<br />

Magnetic impurity phases are not important.<br />

On the other hand, if I have a pure sample can determ<strong>in</strong>e<br />

NMR<br />

pauli<br />

macroscopic<br />

<br />

<br />

pauli<br />

N<br />

0 2<br />

Examples: <strong>Knight</strong> <strong>Shift</strong> measurements <strong>in</strong> superconductors<br />

B<br />

B 0<br />

SC<br />

normal metal<br />

T c<br />

T


(4)<br />

Superconduct<strong>in</strong>g state below T C<br />

All electrons condense <strong>in</strong>to a s<strong>in</strong>gle state described by a macroscopic wave function.<br />

i r<br />

r r e <br />

<br />

0<br />

<br />

Fermi Dirac statistics does not apply!<br />

electrons form pairs that are described by Bose statistics<br />

Cooper-pairs, mediated by electron-phonon <strong>in</strong>teraction<br />

Concern<strong>in</strong>g the sp<strong>in</strong> there are two possibilities:<br />

S 0<br />

<strong>in</strong> most cases<br />

S 1<br />

sp<strong>in</strong>-part<br />

odd<br />

even<br />

orbital part of wavefunction<br />

s- d- wave<br />

L 0,2,...<br />

even<br />

with respect to <strong>in</strong>version<br />

p- f- wave<br />

L 1,3,...<br />

odd<br />

<strong>in</strong> most superconductors, especially the classical ones:<br />

SC<br />

pauli 0 at T 0<br />

SC<br />

<br />

2<br />

density of the SC condensate<br />

T<br />

K<br />

T <strong>in</strong>dependent <strong>in</strong> the normal state D E<br />

<br />

F<br />

<br />

T c<br />

T<br />

K<br />

0 for T 0<br />

<br />

pairs


(5)<br />

We can also learn about the symmetry of the orbital part of the wave function,<br />

s-wave or d-wave:<br />

<br />

–<br />

s<br />

wave<br />

+<br />

+<br />

+<br />

+<br />

+<br />

+<br />

–<br />

d<br />

wave<br />

Consider energy gap at the Fermi-surface:<br />

energy required to break up Cooper pairs and create unpaired electrons<br />

s-wave<br />

k y<br />

f<strong>in</strong>ite gap over entire<br />

Fermi surface<br />

k x<br />

d-wave<br />

–<br />

nodes where the gap vanishes<br />

electrons can be excited even at very small energy<br />

+ +<br />


(6)<br />

SC<br />

2<br />

swave<br />

d<br />

wave<br />

T c<br />

T<br />

K<br />

d<br />

wave<br />

s<br />

wave<br />

T<br />

This is the case for high T C superconductors<br />

S 1<br />

for example Sr 2 RuO 4 T C =1.5 K<br />

where superconductivity may be mediated by ferromagnetic fluctuations<br />

K<br />

pwave L<br />

1<br />

T c<br />

d wave L<br />

swave L<br />

2<br />

0<br />

T<br />

– advantage: simple experiment that can be performed on polycrystall<strong>in</strong>e samples<br />

– impurity phases are not contribut<strong>in</strong>g local probe!


(7)<br />

1<br />

Sp<strong>in</strong> lattice relaxation rate <strong>in</strong> metals<br />

T1<br />

due to sp<strong>in</strong>-flip <strong>in</strong>teraction between conduction electrons and nuclear sp<strong>in</strong>s.<br />

i<br />

e<br />

<br />

<br />

k , <br />

<br />

nucleus<br />

f<br />

nucleus<br />

<br />

<br />

k,<br />

<br />

e<br />

<br />

Concern<strong>in</strong>g nucleus, consider<br />

1<br />

I , this corresponds to a transition<br />

2<br />

+ ½<br />

- ½<br />

E g<br />

B<br />

N<br />

N<br />

Transition rate is determ<strong>in</strong>ed by Fermi’s golden rule (2nd order perturbation theory)<br />

2<br />

2<br />

W<br />

f H i E E<br />

<br />

E<br />

<br />

<br />

hf N k k<br />

8<br />

2 <br />

0<br />

SI<br />

3 <br />

with Hhf e e N<br />

hyperf<strong>in</strong>e coupl<strong>in</strong>g<br />

constant a<br />

very<br />

small<br />

<br />

Consider matrix elements fSIi<br />

<br />

i ,<br />

k <br />

<br />

<br />

SI Sz Iz SyIy SxIy<br />

<br />

1<br />

SI <br />

SI <br />

2<br />

1<br />

with Sx<br />

iSy<br />

2<br />

where S , I rais<strong>in</strong>g operator<br />

S<br />

<br />

<br />

, I<br />

<br />

<br />

lower<strong>in</strong>g operator<br />

<br />

f ,<br />

k <br />

<br />

electron part<br />

nuclear


(8)<br />

S<br />

S<br />

<br />

<br />

S or 0<br />

z<br />

<br />

0<br />

1<br />

<br />

, k<br />

IzSz<br />

IS IS<br />

,<br />

k <br />

2 2<br />

8<br />

fH 0<br />

<br />

3<br />

2 2<br />

hf i<br />

el e N<br />

<br />

F<strong>in</strong>ally we need to consider the density of available electronic states.<br />

i must be occupied f must be empty<br />

2<br />

f<br />

f E<br />

kB<br />

T<br />

1<br />

f E<br />

E F <br />

E<br />

only this range of<br />

energies contributes<br />

<br />

1<br />

<br />

W<br />

dE dE D E f E<br />

D E f E<br />

s<strong>in</strong>ce E E<br />

<br />

<br />

<br />

E<br />

EF<br />

<br />

exp<br />

<br />

2<br />

kB<br />

T<br />

1<br />

dE dE D E <br />

with f E<br />

<br />

E EF<br />

E EF<br />

<br />

1exp 1exp <br />

kB<br />

T kB<br />

T<br />

<br />

<br />

f E<br />

kB<br />

T<br />

E<br />

f<br />

s<strong>in</strong>ce dE E E<br />

E<br />

2<br />

W<br />

D E k T<br />

<br />

F<br />

<br />

B<br />

F<br />

<br />

8<br />

2 <br />

consider<strong>in</strong>g the matrix element fHfh i e 0<br />

e N <br />

3 2<br />

2<br />

2<br />

2<br />

8<br />

2 <br />

W <br />

e0 e N DEFkB<br />

T<br />

3 2 <br />

<br />

<br />

<br />

<br />

<br />

density of available<br />

states<br />

2


(9)<br />

1<br />

T<br />

1<br />

W W 2W<br />

<br />

<br />

<br />

2<br />

4<br />

8<br />

2 <br />

e0 e N DEF kB<br />

T<br />

3 2 <br />

<br />

<br />

2<br />

1<br />

constant D EF e<br />

T T<br />

1<br />

0 4<br />

Korr<strong>in</strong>ga-relationship<br />

8<br />

with K <br />

3<br />

<br />

e<br />

0<br />

N<br />

2<br />

<br />

pauli<br />

<br />

2<br />

DE <br />

B<br />

F<br />

<br />

1 e<br />

1<br />

<br />

T T 4k K<br />

1<br />

B<br />

N<br />

2<br />

<strong>Knight</strong> shift<br />

Experiment:<br />

1 measurement on YBa Cu O<br />

T<br />

1<br />

2 3 7<br />

YBa 2 Cu 3 O 7


(10)<br />

1<br />

measured –relaxation rate on Cu and O site<br />

T<br />

1<br />

The relaxation rates at all four oxygen sites and<br />

both Cu sites <strong>in</strong> a 7.0T ( 17 O) or 7.4T ( 63 Cu) field<br />

applied along the c-axis of the oriented powder,<br />

versus temperature, plotted on a log-log scale.<br />

The solid l<strong>in</strong>es <strong>in</strong>dicate a l<strong>in</strong>ear relationship<br />

between relaxation rate and temperature. The<br />

<strong>in</strong>set emphasizes the l<strong>in</strong>ear temperature<br />

dependence of oxygen relaxation above T C .<br />

P.C. Hammel et al., Phys. Rev. Lett. 63 (1989),<br />

1992.<br />

Cu<br />

O<br />

Cu<br />

YBa 2<br />

Cu 3<br />

O 6<br />

AF correlated Cu-sp<strong>in</strong>s<br />

2<br />

2<br />

9<br />

Cu 3d shell, 1 electron miss<strong>in</strong>g S<br />

O<br />

full p-shell<br />

S<br />

O<br />

0<br />

Cu<br />

1<br />

<br />

2<br />

Korr<strong>in</strong>ga relation is fulfilled for O NMR signal, but violated for Cu NMR signal.<br />

Short range AF magnetic correlations of Cu moments persist <strong>in</strong> the<br />

superconduct<strong>in</strong>g state!<br />

Many open questions<br />

Once more it is very important to dist<strong>in</strong>guish <strong>in</strong>tr<strong>in</strong>sic signal from extr<strong>in</strong>sic (i.e.<br />

from impurity phases).

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!