Physikalische
Folge,
164,
(1989)
©
by
R.Oldenbourg
Verlag,
Miinchen 1989
-0044-3336/89
$3.00+0.00jjSR Investigations
of Cerium
Hydride*
P.
Birrer,
F.N.Gygax,
B.Hitti,
E.Lippelt
and A. SchenckInstituteforIntermediate
Energy Physics
(IMP),
ETHZ,
CH-5234Villigen,
SwitzerlandL.
Schlapbach
Institut de
Physique,
Universite deFribourg,
CH-1700Fribourg,
SwitzerlandThe
Knight
shift and transversefieldspin
relaxation rateofpositive
muons(/x+),
implanted
inLaH2.75-,
CeH2.7o-
andCeH2.95-powder samples,
have been measuredas afunction of
temperature
andin theCerl^-samples
also as afunction offield.These data are used to extract informationon the
/i+-site, ^-diffusion
and themagnetic properties
of theCeH^-samples.
Introduction
Cerium
hydrides
CeHj.
areknowntoexhibitfascinating
properties
regarding
theirphase
diagram j\j.
Their electronic structure/2/
changes
from metallicat x =2to
insulating
(or
semiconducting)
at x = 3 and theirmagnetic properties
includeboth
ferromagnetic
order(for
2.3 < x <2.7)
andantiferromagnetic
order(for
x < 1.95 and x >2.75) /3/.
In addition measurements ofthe electronicspecific
heat ina
CeH2
65sample yielded
evidence forheavy
fermion behavior/4/.
Hydro-gen
occupies
up to x = 2the available tetrahedral sites in theCe
fee lattice.On
1048
P. Birreretal.The indication for
heavy
fermion behavior and thecomplex magnetic phase
dia-gram in these
systems
motivated us toinvestigate
them furtherby
means of thefiSR. technique
/5/
which has proven to be of considerablepower in thestudy
ofheavy
fermionsystems
(see
e.g./6/).
In addition we wereinterested in thediffu-sional behaviorof the
fi+,
inits site of residence(tetrahedral
oroctahedral)
andits interaction with the
neighboring
protons.
In this contribution we
report
on first results from asystematic
/xSR
study
ofcerium
hydrides
/7/.
Twopowder
samples
with nominal values ofx = 2.70 and x =2.95 which bracketed well the metal tosemiconductor transitionwereinvesti-gated.
Thesample
with x =2.95 shouldshowantiferromagnetism
withTjv
—
4K while the other
sample
wasexpected
to be eitherferromagnetic
withTc
around~ 2 K or tofall within the narrow
paramagnetic region
abovex = 2.70/3/.
Forcomparison
also results fromanonmagnetic
LaH2
75sample
areincluded here. Theexperiments
wereperformed
at the Paul Scherrer Institute(formerly SIN).
Resultsand discussion
Generally
the/xSR signal
in theCeHj.
compounds
consisted oftwocomponents,
irrespective
of theapplied
field. Theamplitudes
ofthe twocomponents
showeda weak
temperature
dependence
and astronger
fielddependence,
however their sumremainedindependent
oftemperature
orapplied
field. Thisimplies
that themuons do not
probe
two differentphases
in thesamples
but that the twocom-ponents
areprobably
originating
fromfi+
in differentmagnetic
environments theextent of which is controlled
by
applied
field andtemperature.
The two compo-nentsaredistinctby
their relaxationratesandfrequencies.
At afield of0.02 Ttheminority
component
(called
the fastcomponent)
displayed
a rathertemperature
independent
relaxation rate of~ 1.25fis'1
for bothCeHx
samples.
Thiscompo-nent
disappeared
close to roomtemperature.
At lowtemperatures
theminority
component
(~ 25%)
became themajority
component
(~
75%)
forapplied
fields above ~ 0.08 T. Above ~ 0.1 T the relaxation rate increasedlinearly
with thefield. The other
component
(called
the slowcomponent)
displayed always
amuchsmaller relaxationratewitha
pronounced
temperature
dependence
butnofieldde-pendence.
Thissignal
resembles thesingle
component
found in theLaH2.75
sample.
;jSR
Investigations
Hydride
The
magnitude
of the relaxationrateof the fast componentcanonly
beexplained
if static electronic moments are invoked. Since we have no clear idea about the
origin
ofthiscomponent,
wewill discusshereonly
the slowcomponent
whichhasmanyfeatures incommon with proton NMR
signals
inLaHj.
andCeHx
/8,9,10/.
Fig.
lashows thetemperature
dependence
ofthefi+
relaxation rate orlinewidth orinLaH2.75,
displaying
the onset of motionalnarrowing
around 50 Kand,
morepronounced,
above ~ 150 K. Around the lattertemperature
hydrogen
diffusionisknownto setin
/8/.
Similarcurves wereobtainedinCeH2.7o
(Fig. lb)
andCeH2.95
(Fig. lc).
The lowtemperature
plateau
values of or can be used to learn about the/x+
position
in thelattice. In Tab. 1 wecomparetheexperimental
valueswith calculatedpredictions
for the twodifferentpossible
sites,
assuming
forthe lattice constant a„ =5.57A.
Wesee thatin all cases theexperimental
values aresignifi-cantly larger
than thecalculatedones. Toobtainagreement
wecan scaledown thelattice constants
(<r
oca~3).
Foranoctahedralsiteassignment
this wouldimply
asurprisingly
large
nearneighbor
latticecontractionof—8.5%
inCeH2.70,
—5.5%
inCeH2.95
and—3.2%
inLaH2
76. Thesevalueswould beevenlarger
foratetrahedralsite
assignment
which seems to ruleout the latter one. InCerl*
the lowtemper-ature <tmay be enhanced
by magnetic
effects,
nevertheless it appears as if thereis a
strong
attractiveinteraction betweenthefi+
at an octahedral interstitial siteand the nearest
neighbor
protons
at tetrahedral interstitial sites.<texp(/is
')
<rcal(a0
=5.57A)
a„,e//
(A)
aocxpA
An/a
(%)
octh. tetr. octh. octh.
LaH2
75 0.26 0.244 0.220 5.450 5.63/8/
3.2CeH2.70
0.31 0.233 0.196 5.064 5.54/ll/
-8.5CeH2.9B
0.28 0.235 0.212 5.254 5.54/ll/
-5.5 Tab. 1.Compilation
of resultson cr andcomparison
withcalculations. Thetemperature
dependence
of<r athigher
temperatures
wasused to extract thefi+
diffusionrate ormoreprecise,
acorrelationratev= +up whichcontainsalsothe
proton
jump
rateup. It is found thatv(T)
followedanArrhenius law fortemperatures
above about 180 K. Tab. 2 containsthefitted
activationenergies
andpreexponential
factorstogether
withcorresponding
data
fromaproton
NMR1050
P. Birreretal. La HSS
0.20 s oj & 0.10 0.00 L-LlJ-1_L_ 10u 10' Temperature(K ) CeH,05 0.40101
Temperature(K)Fig.
1.Temperature dependence
of (t^out m(a)
LaH2.7o,
(b)
CeH2.7o
and(c)
CeH2.95-
Thedatawereobtainedin an
applied
field of0.02 T.study.
It is seen thatthe
present
activationener-gies
andpreexponential
fac-tors are
significantly
smallerthan in the
proton
case. Asimilar observation was also
made in
PdH*
/12/.
This is hard tounderstand,
be-cause one
expects
always
v >up. A way out would be
a
highly
correlatedfi+-proton
diffusion or a kind of
immo-bilizationofthe
tightly
boundft+
—
nn
proton
complex.
An-other
possibility
is astoichio-metric
CeH3
environment in thefi+
vicinity
which should also inhibitstrongly
uncorre-lated
jump
events.The
Knight
shift of thep+
Larmorfrequency
inLaH2.75
was too small to be detected in the
present
study.
Incontrast the
Knight
shift in theCeHj.-samples
wasquite
sizeable and showed remark-able
temperature
dependen-cies
(see
Fig.
2a,b).
InCeH2.7o
the
Knight
shiftchanges
from about—0.2%
at roomtem-perature
to about+0.25%
at~ 5
K,
displaying
a shallowmaximum there. In contrast
uSR
Investigations
CeriumHydride
fi+
E(kcal/Mol)
f„(a_1)
LaH2
75 - 2.28 1.2•10*
CeH2.7o
~0.50 4.6 •106
CeH2
95 ~1.63 3.2-107
PLaH2.86
~3.50 ~10"
/8/
Tab. 2. Diffusion
parameter
for T > 180 K.CeH 2.778 2.772 2.766 2.760 10 10' Temperature(K) CeH,0c 2.778 2.772 2.766 2.760 tt-1-r 10" 101 Temperature IK I —1_i_.1_i_l
Fig.
2.Temperature dependence
of/z+-Larmor
frequency
of the slowcomponent
in(a)
CeH2.7o
and(b)
CeH2.95.
The dashed line
corresponds
to the external field(~
0.02
T).
close to zero both at room
temperature
and below 5 K and has a shallow minimumof ~
-0.2%
at 30 K. Afew
proton
Knight
shift data forCeHj.
with x = 2.85 canbe found in
/13,9/
fortem-peratures
between 6 K and 180 A".They
agreerelatively
well with thepresent
results inCeH2.70.
Themagnitude
of theKnight
shift inFig.
2clearly
points
to astrong
in-fluence of the Ce4f-moments.
The
negative
shift inCeH2.70
at roomtemperature
seemsto
suggest
anantiferromag-netic
coupling
ofthe conduc-tion electrons(or
what ever screens the/i+-charge)
to the 4f-electrons. Thechange
ofsign
at ~ 20 K is for nowa
mystery
as is the differencebetween the results in
Figs.
2a and 2b.1052
P. Birreretal. ConclusionsPositive muons
implanted
inCeH,,.
(x
=2.70,
2.95)
andLaH2.76
seem to occupythe octahedral interstitial sites. The nearest
neighbor
protons
at tetrahedral in-terstitial sites arecontracted towards the/i+
by
afew%.
Uncorrelated diffusionalmotion ofthe
fi+
and itsproton
neighbors
is slowed downconsiderably compared
toproton
diffusion in the bulk ofthe material. The/i+
Knight
shift showsquite
a different andunusual
temperature
dependence
for the twoCeHj.
samples.
It issuggested
that thehydrogen
concentration via thedensity
of free carriers has apronounced
influenceonthe(exchange)
coupling
of the electronsscreening
thefi+
(mostly
conductionelectrons?)
and the 4f-electrons. Whether this is correlated with the metal-semiconductortransition remains tobeseen. The lowtemperature
magnetic phase
transitionwas not reflectedin the/xSR data, although
somekindof weak static
magnetism
alsoathigher
temperatures
wasexperienced by
part
ofthe
implanted
/x+.
References