Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
M.V. Romalis and M. P. Ledbetter- Transverse Spin Relaxation in Liquid ^129-Xe in the Presence of Large Dipolar Fields

M.V. Romalis and M. P. Ledbetter- Transverse Spin Relaxation in Liquid ^129-Xe in the Presence of Large Dipolar Fields

Ratings: (0)|Views: 15|Likes:
Published by Gravvol

More info:

Published by: Gravvol on Feb 17, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

10/30/2013

pdf

text

original

 
V
OLUME
87, N
UMBER
6 P HYS ICAL REVIEW LETTER S 6 A
UGUST
2001
Transverse Spin Relaxation in Liquid
129
Xe
in the Presence of Large Dipolar Fields
M.V. Romalis and M.P. Ledbetter
 Department of Physics, University of Washington, Seattle, Washington 98195
(Received 5 March 2001; published 23 July 2001)Using spin-echo NMR techniques we study the transverse spin relaxation of hyperpolarized liquid
129
Xe
in a spherical cell. We observe an instability of the transverse magnetization due to dipolar fieldsproduced by liquid
129
Xe
, and find that imperfections in the
pulses of the spin-echo sequence suppressthis instability. A simple perturbative model of this effect is in good agreement with the data. We obtaina transverse spin relaxation time of 1300 sec in liquid
129
Xe
, and discuss applications of hyperpolarizedliquid
129
Xe
as a sensitive magnetic gradiometer and for a permanent electric dipole moment search.
DOI: 10.1103/PhysRevLett.87.067601 PACS numbers: 76.60.Jx, 07.55.Ge, 33.25.+k 
Atomic magnetometers are widely used in precisionmeasurements, such as searches for a permanent electricdipole moment (EDM) [1,2] and CPT violation [3]. Thefundamental sensitivity limit of an atomic magnetometer isdetermined by the uncertainty in the measurement of theZeeman precession frequency
vdv
1
NT 
2
t
,
(1)where
is the number of atoms,
2
is the transverse spinrelaxation time, and
t
is the measurement time. Mag-netometers using electron spin precession in Rb atoms[4] and nuclear spin precession in
199
Hg
atoms [2] haveapproached this fundamental limit of sensitivity. Liquid
129
Xe
is a very attractive substance for a magnetometerbecause of its high number density
10
22
cm
3
and po-tentially long values of 
2
. It also has a high electric
eld breakdown strength of 
400
kV
cm [5], making itan excellent candidate for an EDM search. Large quan-tities of hyperpolarized liquid
129
Xe
have recently beenproduced using spin exchange with optically pumped Rbatoms [6].In this Letter we describe our experimental and theo-retical studies of the transverse spin relaxation of liquid
129
Xe
. We describe new nonlinear effects in the relaxationof the transverse magnetization due to large dipolar mag-netic
elds created by hyperpolarized
129
Xe
. The effectsof dipolar
elds in NMR have recently attracted much at-tention. They have been studied using liquid
129
Xe
[7] and
3
He
[8] in aU-shaped sample tube. Theoretical models forthese effects were considered in [9]. Unlike previous theo-retical and experimental studies, we work in a sphericalcell geometry and develop a simple perturbative model todescribe the nonlinear effects in the regime of weak mag-netic
eld gradients, when the magnetization of 
129
Xe
re-mains nearly the same everywhere in the cell. We
nd thatunder these conditions the transverse magnetization is un-stable and small gradients ofthe external magnetic
eldareexponentially ampli
ed. Similar behavior was observed ina simple numerical model consisting of eight spins locatedat the corners of a cube [9]. We also
nd that imperfec-tions of 
pulses in a standard Carr-Purcell-Meiboom-Gill(CPMG) spin-echo sequence [10] suppress the exponentialgrowth of the magnetization gradients. Our experimentalresults are in good quantitative agreement with the pertur-bative model. After the instability is suppressed, we obtain
2
1300
sec in liquid
129
Xe
, which, we believe, is thelongest transverse spin relaxation time measured in a liq-uid. We also discuss how the exponential ampli
cation of the magnetic
eld gradients caused by dipolar interactionsin liquid
129
Xe
can be used to build a very sensitive mag-netic gradiometer and to search for an EDM. For theseapplications it is convenient to use a SQUID magnetome-ter to detect
129
Xe
magnetization with high signal-to-noiseratio in a very low magnetic
eld.The longitudinal spin relaxation time
1
of liquid
129
Xe
has beeninvestigated in [6,11]; at180 Kit is about 30 min.Spin relaxation in liquid
129
Xe
is dominated by spin-rotation interactions, and an estimate of 
1
from the chem-ical shift of 
129
Xe
[12] is in good agreement with theexperiment. In a liquid one might expect that the correla-tion time
t
c
of the spin-rotation interaction would be veryshort, and in the regime of motional narrowing
t
c
v
ø
1
2
should be equal to
1
. However, existence of long-livedXe van der Waals molecules [13,14] could result in shortervalues of 
2
. In previous studies of the transverse relax-ation in liquid
129
Xe
[15,16] measured values of 
2
havebeen no longer than several seconds.Our measurements are performed using
129
Xe
polarizedby spin-exchange optical pumping [17]. A schematic of the apparatus is shown in Fig. 1. A mixture of 2% Xe(in natural abundance), 2%
N
2
, and 96%
4
He
ows at apressure of 3 atm through an optical pumping cell con-taining Rb vapor at 155
±
C. The Rb is optically pumpedby a 40 W diode laser array whose spectrum is narrowedusing an etalon [18]. Polarized
129
Xe
gas freezes out of the mixture as it passes through a cold trap (77 K) in amagnetic
eld of 1.5 kG, where it has a longitudinal spinrelaxation time of several hours. After 40 min of accumu-lation, approximately 7 g of Xe ice is collected in the coldtrap. The trap is then warmed up and Xe gas
ows to a1.3 cm diam. spherical glass cell, where it lique
es at atemperature of 180 K maintained by a mixture of acetoneice and liquid.067601-1 0031-9007
01
87(6)
067601(4)$15.00 © 2001 The American Physical Society 067601-1
 
V
OLUME
87, N
UMBER
6 P HYS ICAL REVIEW LETTERS 6 A
UGUST
2001
FIG. 1. Schematic of the apparatus. The magnetic
eld, cre-ated by four coils, has two uniform regions near the pumpingcell and the liquid Xe cell.
The NMR measurements are performed in a magnetic
eld of 32 G
v
2
p 3
37.5
kHz
. To eliminateradiation-damping effects the NMR coil is connectedto a high-impedance ampli
er without using a resonantcircuit. The gradients of the external magnetic
eld arereduced to about
0.1
mG
cm by
rst-order gradient coils,so that the free-induction decay time due to dephasing of the spins is
2
2 3
sec. The timing of the RF pulsesis controlled by an NMR pulser PC card [19], while thephases of the pulses are set with high precision using aDSP function generator that is controlled through serialinterface commands also generated by the pulser card.Our measurements of 
1
are in agreement with [6].Tosuppress spindephasingdue toresidual externalmag-netic
eld gradients we use a standard CPMG spin-echopulse sequence [10]. It consists of a
2
pulse followedby a train of 
pulses at times
t
,3
t
,5
t
,...
, whose phasesare shifted by 90
±
from the phase of the
2
pulse. Theduration of the
pulses is about 1 msec. Typical values of 
t
are 30 to 100 msec, much shorter than
2
. The decay of the magnetization due to spin diffusion between
pulsesis negligible.The spin-echo technique does not prevent spin dephas-ing due to gradients created by the dipolar
elds, sincethese gradients are reversed by
pulses together withthe magnetization. For a uniform
129
Xe
polarization ina spherical cell the dipolar
eld seen by
129
Xe
atoms addsup to zero. However, in the presence of a small gradient of the external magnetic
eld the magnetization of 
129
Xe
willdevelop a helix which in turn produces a gradient of themagnetic
eld. We
nd that this positive feedback mecha-nism causes the gradients of the magnetic
eld and themagnetization to grow exponentially in time and results ina highly nonexponential decay of the transverse magneti-zation shown in Fig. 2.
0 20 40 60 80 100
Time (sec)
051015
    T   r   a   n   s   v   e   r   s   e    M   a   g   n   e    t    i   z   a    t    i   o   n    (
     µ
    G    )
FIG. 2. Onset of nonexponential decay due to dipolar
elds.Open circles: envelope of the transverse magnetization signalobtained with CPMG pulse sequence for
t
100
msec andan applied longitudinal
eld gradient
dH 
Ez
dz
1.4
mG
cm.Solid line: model of the initial magnetization gradient growthwith no free parameters.
We develop a simple model for a spherical cell to cal-culate the rate of exponential growth of the gradients andalso incorporate into the model the effects of imperfect
pulses. The magnetic
eld felt by
129
Xe
atoms is
B
H
E
1
H
1
4
3
M
,
(2)where
H
E
is the external magnetic
eld and
H
is the
eldcreated by
129
Xe
magnetization [20]. This expression dif-fers from the classical result
B
H
1
4
M
because two
129
Xe
atoms cannot occupy the same space and the
d
func-tion part of the classical dipolar
eld does not contribute.If the initial
2
pulse is applied along the
ˆ
directionin the rotating frame, the magnetization is rotated into the
ˆ
x
direction. Before the magnetization is signi
cantly af-fected by the
eld gradients, only two terms of the Blochequations are signi
cant,
y
t
2g
0
B
z
and
z
t
g
0
B
y
.
(3)Only the gradients of 
Ez
are important, since the gradi-ents of 
Ey
will average to zero in the rotating frame. We
rst consider the case when the longitudinal gradient of 
Ez
dominates,
H
E
gz
ˆ
z
. To
rst order the magnetiza-tion will also develop a linear gradient in the
ˆ
z
direction,and we look for solutions in the form
y
r
,
t
m
y
t
z
R
,
z
r
,
t
m
z
t
z
R
, where
R
is theradius of the cell. The total magnetic
eld is
B
y
8
15
m
y
t
zR
and
B
z
2
16
15
m
z
t
zR
1
gz
.
(4)In Eqs. (4) we omitted
y
gradients of the magnetic
eldbecause their effect averages to zero in the rotating frame.Equations (3) and (4) have a simple solution,067601-2 067601-2
 
V
OLUME
87, N
UMBER
6 P HYS ICAL REVIEW LETTERS 6 A
UGUST
2001
m
y
t
2g
0
gR
sinh
b
t
b
,
(5)where
b
8
2
g
0
k
15
and
k
k
L
1
for the lon-gitudinal gradient of 
Ez
. For transverse gradients of 
Ez
in the
x
or
y
direction of the lab frame the magnetizationdevelops a gradient in the same direction. The solutionis still given by Eq. (5) with
k
k
1
2
for
b
ø
v
.Thus, the gradients of the magnetization grow exponen-tially with a time constant that in our experiment is on theorder of 2
20 sec. Even for a very small external mag-netic
eld gradient,
m
y
0
will quickly approach unity, atwhich point the transverse magnetization will dephase ona time scale of 
b
2
1
due to its own
eld gradients.In the CPMG pulse sequence the magnetization and itsgradients are reversed at times
t
,3
t
,5
t
,. ..
. This can berepresented in Eqs. (4) by an alternating external
eld gra-dient
g
. Inthis case thedifferential equations(3) are solvednumerically, but for
tb
ø
1
and
t
¿
t
the numerical so-lution is well approximated by
m
y
t
2
2
2
pk
15
g
2
20
t
2
gRe
b
t
.
(6)The gradient of the magnetization that appears before the
rst
pulse continues to grow exponentially and thespin-echo envelope starts to deviate from initial exponen-tial decay at a time
t
ne
when
m
y
t
ne
0
. For param-eters corresponding to the data shown in Fig. 2 we obtain
b
2
1
5
sec, and
t
ne
20
sec, in good agreement withthe data given exponential sensitivity to input parameters.Equation (6) suggests that even for a very small mag-netic
eld gradient and
pulse spacing the magnetizationwill eventually decay nonexponentially after a time on theorder of several
b
2
1
. However, as shown in Fig. 3, thegrowth of the magnetization gradients can be suppressedby imperfections in the
pulses of the CPMG sequence.Since the
pulses are applied along the
ˆ
x
direction in therotating frame, the net effect of an inaccurate pulse lengthis to slowly rotate
y
and
z
components of the magne-tization around the
ˆ
x
direction with an angular frequency
v
r
f

4
pt
, where
f
a 2
, and
a
is the truelength of the
pulse. Adding this rotation to Eqs. (3),one
nds
b
∑µ
16
15
k
2
g
0
2 v
r
µ
8
15
g
0
1 v
r
∂∏
1
2
,
(7)where
k
is equal to
k
L
or
k
depending on the direc-tion of the external
eld gradient. Thus,
b
is imaginaryfor
v
r
.
16
15
k
2
g
0
or
v
r
, 2
8
15
g
0
and themagnetization gradients do not grow exponentially. Thesolid points in Fig. 3 show the spin-echo envelope with
pulses intentionally shortened by 3%, which delaysthe onset of the nonexponential decay to about 250 sec.For the parameters corresponding to these data we
nd
v
r
2
0.09
sec
2
1
and
8
15
g
0
0.1
sec
2
1
, so
b
given by Eq. (7) is close to zero within errors.Figure 3 also shows the spin-echo envelope for a CPMGsequence with a smaller
0
and a shorter
pulse spac-
0 1000 2000 3000 4000
Time (sec)
0246810
    T   r   a   n   s   v   e   r   s   e    M   a   g   n   e    t    i   z   a    t    i   o   n    (
     µ
    G    )
FIG. 3. Spin-echo envelope for CPMG sequence with
t
100
msec,
dH 
Ez
dx
1.4
mG
cm, and the
pulse intention-ally shortened by 3% (solid circles); with
t
30
msec andgradients reduced to about
0.1
mG
cm (open circles). An ex-ponential
t to the latter gives
2
1290
sec (solid line). Fitsto the initial and
nal decay of the data shown with solid circlesalso give similar values of 
2
.
ing
t
, which increases
v
r
. In this case the transversemagnetization decays exponentially, with a time constant
2
1290
sec. In other data sets we obtained values of 
2
up to 1360 sec. Most systematic effects would de-crease the value of 
2
, and we cannot exclude the pos-sibility of an additional relaxation rate on the order of 
1
2
2
1
1
2
3
10
2
4
sec
2
1
. For example, it couldbe due to partial transverse spin relaxation on the walls of the cell. Note that if the walls of the cell were completelydepolarizing, it would give a diffusion-limited relaxationrate of 
1
3
10
2
3
sec
2
1
. Thus, our data are also consistentwith intrinsic
2
for liquid
129
Xe
equal to
1
1800
sec.Using our measurements of 
2
, the fundamental limiton the sensitivity of a liquid
129
Xe
magnetometer, givenby Eq. (1), is estimated to be
3
3
10
2
17
G
Hz, 5 ordersof magnitude better than the best SQUID magnetometers.However, to realize this sensitivity one needs an ef 
cientway of detecting the frequency of precession of 
129
Xe
spins. A SQUID magnetometer can detect the precessingmagnetic
eld created by polarized
129
Xe
with a highsignal-to-noise ratio. The magnetic
eld created by
129
Xe
in natural abundance (26%) with 20% spin polarizationcontained in a 1 cm diameter cell is about 5 mG. Fora SQUID sensitivity of 
10
2
11
G
Hz one would geta signal-to-noise ratio of 
5
3
10
8
Hz. Such highsignal-to-noise can be realized in SQUIDs using
uxcounting techniques [21].The dipolar self-interaction effects discussed above alsooffer a very attractive technique for using liquid
129
Xe
asa sensitive gradiometer. According to Eq. (5) the magneti-zation gradients produced by an external
eld gradient areexponentially ampli
ed with a time constant
b
2
1
. Liquid
129
Xe
essentially acts asits own magnetometer, amplifying067601-3 067601-3

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->