Quasiparticle relaxation dynamics in spin-density-wave and superconducting SmFeAsO_{1-x}F_{x} single crystals

2 MB·English

by  T. Mertelj

Quasiparticle relaxation dynamics in spin-density-wave and superconducting
SmFeAsO F single crystals
1−x x
T. Mertelj1, P. Kusar1, V.V. Kabanov1, L. Stojchevska1, N.D. Zhigadlo2, S.
Katrych2, Z. Bukowski2, J. Karpinski2, S. Weyeneth3 and D. Mihailovic1
1Complex Matter Dept., Jozef Stefan Institute, Jamova 39, Ljubljana, SI-1000, Ljubljana, Slovenia
2Laboratory for Solid State Physics, ETH Zürich, 8093 Zürich, Switzerland and
3Physik-Institut der Universität Zürich, 8057 Zürich, Switzerland
(Dated: January 8, 2010)
0
We investigate the quasiparticle relaxation and low-energy electronic structure in undoped Sm-
1
0 FeAsO and near-optimally doped SmFeAsO0.8F0.2 single crystals - exhibiting spin-density wave
(SDW)ordering and superconductivity respectively - using pump-probefemtosecond spectroscopy.
2
Intheundopedsingle crystalsasinglerelaxation process isobserved,showing aremarkablecritical
n slowingdownoftheQPrelaxationdynamicsattheSDWtransitiontemperatureTSDW ≃125K. In
a
thesuperconducting(SC) crystals multiplerelaxation processes are present,with distinct SC state
J
quasiparticlerecombinationdynamicsexhibitingaBCS-likeT-dependentsuperconductinggap,and
8 a pseudogap (PG)-like feature with an onset above 180K indicating the existence of a pseudogap
of magnitude 2∆PG ≃120 meV above Tc. From the pump-photon energy dependence we conclude
]
that the SC state and PG relaxation channels are independent,implying the presence of two sepa-
n
o rate electronic subsystems. We discuss the data in terms of spatial inhomogeneity and multi-band
c scenarios, findingthat the latter is more consistent with thepresent data.
-
r
p
The discovery of high-temperature superconductivity tronic structure, investigating possible multi-component
u
s iniron-basedpnictides (IP)1–3 hasattractedagreatdeal response as a sign of phase separation and to obtain de-
. of attention recently. Contrary to the cuprate supercon- tailed information about the quasiparticle (QP) dynam-
t
a ductors,whereasinglebandwithahighdegreeofcorre- ics in the normal, SDW and superconducting states.
m
lations is believed to be sufficient starting point for the
- description of the electronic properties, there is a clear
d
theoretical4 and experimental5 evidence that in IP sev-
n I. EXPERIMENTAL
eral bands cross the Fermi energy (ǫ ).58The implica-
o F
c tions ofthe presenceofseveralbandsat ǫF inIParestill
Optical experiments were performed using the stan-
[ under intense investigation. In the undoped state two
SDWgapsweredetectedbyopticalspectroscopy6 in122 dard pump-probe technique, with 50 fs optical pulses
2
from a 250-kHz Ti:Al O regenerative amplifier seeded
v compounds (AFe2As2 A=Ba,Sr) presumably originating 2 3
with an Ti:Al O oscillator. We used the pump pho-
47 afrosimmidlaiffreorpentitcablacnodnsdcurcotsivsiintygsǫuFp.pIrnesLsaioFneAwsaOso(bLsae-r1v1e1d17), tons with eithe2r 3doubled (~ωP = 3.1 eV) or fundamen-
0 but no analysis in terms of SDW gaps was performed. tal (~ωP = 1.55 eV) photon energy and the probe pho-
1 In the superconducting state multiple superconducting tons with 1.55 eV photon energy. The pump and probe
1. gaps were detected8–11 corresponding to different bands polarizations were perpendicular to each other and ori-
entedwithrespecttothe the crystalsto obtainthe max-
0 crossingǫ . Inadditiontosuperconductinggapsalsothe
F
0 presence of a pseudogap was reported by NMR12,13 and imum amplitude of the response at low temperatures.
1 point-contact Andreev spectroscopy14 in La-1111. The pump and probe beam diameters were determined
: by measuring the transmittance of calibrated pinholes
v Time resolved spectroscopy has been very instrumen-
i mounted at the sample place31.
X tal in elucidating the nature of the electronic excitations
The crystals were flux grownat high pressure at ETH
in superconductors, particularly cuprates, by virtue of
r in Zurich32 and were approximately 120 x 80 µm in
a the fact that different components in the low-energy ex- ∼ ∼
size. For optical measurements the crystals were glued
citation spectrum could be distinguished by their differ-
on a sapphire window mounted in an optical liquid-He
ent lifetimes15–29. Moreover, the relaxation kinetics can
flow cryostat.
give us valuable information on the electronic density of
states16 and electron-phonon coupling30. Extensive and
systematicexperimentsoncuprateshavealsogiveninfor-
mationonthebehaviorofthepseudogapforchargeexci- A. Undoped, spin-density-wave ordered SmFeAsO
tations,complementingtheinformationobtainedonspin
excitations from NMR and other spectroscopies17,19,25. In Fig. 1 we plot temperature dependence of ∆R/R
In this work we present a time-resolved femtosec- transients in undoped SmFeAsO. The only discernible
ond spectroscopy study of undoped and near-optimally differenceoftheresponseatdifferentpump-photonener-
doped SmFeAsO1−xFx single crystals with x = 0 and giesisthe presenceofacoherentphononoscillationwith
x 0.2 with the aim of elucidating the low energy elec- thefrequency5.1THz(170cm−1)at295K,at~ω =1.55
P
≃
2
eV, whichis absentat~ω =3.1eV, consistentwith Ra-
P
man data33. Apart of the coherent phonon oscillation
the transients consist from a negative-amplitude single-
exponential relaxation with a temperature independent
risetime of 180fs (see Fig. 2(a)). Around 125Kan
∼ ∼
additional long lived response appears with decay time
beyond our measurement delay range. The amplitude of
thetransients,A ,linearlyincreaseswithdecreasingtem-
0
peraturedownto 170K(seeFig. 2(c))withtherelax-
∼
ationtime, τ , of 220fs time being virtuallyconstant
ud
∼
above200K.Below200Kτ startstoincreasewhilethe
ud
amplitude starts to depart from the linear dependence
only below 170 K rapidly increasing below 140 K,
∼ ∼
and achieving a maximum at 115 K, upon entering the
SDW state. With further decrease of the temperature
the amplitude slightly drops at first and then remains
constantbelow50K.Simultaneouslywiththe maximum
of the amplitude τ shows a remarkable divergent-like
ud
peak at 120 K and then drops to a temperature inde-
∼
pendent value of 0.8 ps below 50 K.
The rise and decay times are virtually independent of
the fluence, ,at alltemperatures(see Fig. 3) while the
F
amplitude increases linearly with at 295K and shows
F
a weak saturation above =25 µJ/cm2 at 5K.
F
Figure 2: ∆R/R transients at representative temperatures
in undoped SmFeAsO with single-exponential decay fits (a).
Therelaxationtimeattwopumpphotonenergiesb)andam-
plitude (c) as functions of temperature at F = 15 µJ/cm2.
The red solid line in (b) is fit of equation (1) to τud above
230K.Thebluedashedlinein(b)isequation(2)withλ=0.2
andΘD =175K.Blackthinsolid linein(a)representsthefit
of equation (28) from16discussed in detail in text. Thin lines
in (b) represent the fits of equation (6) from16 with different
magnitudes of thegap.
the~ω onecanclearlyresolvethreetemperatureregions
P
with different characteristic behaviors.
(i) At the high temperatures a negative transient is
observed with initial 0.25-ps decay followed by a slower
response consisting from a weak peak at 12 ps (see Fig.
5(a))at~ω =3.1eV.Thetransientslinearlyscalewith
P
increasingfluence exceptinthe regionofthe initial0.25-
Figure 1: ∆R/R transients as a function of temperature at
1p.u5m5pe-Vphpoutmonp-epnheortgoynaennderg1y5aµnJd/c1m82µJ(/bc)mi2n(au)ndaonpded3.1SmeV- pFs.dAecta~yωwPh=er3e.1a ewVeatkheFt-rdaenpseinendtesncheaviseothbesesravmedeasthalopwe
and similar amplitude as in undoped SmFeAsO without
FeAsO. In (a) the coherent phonon, shown expanded in the
insert, is artificially smeared beyond 4 ps delay due to a de- the coherent phonon. At ~ωP = 1.55 eV the negative
creased time resolution of scans. hight-T transientsaremuchweakerthanat~ωP =3.1eV
soonlytheinitial0.25-psdecayisresolvedfromthenoise
(see Fig. 6). In addition a coherent phonon is observed
withasofterfrequencythaninundopedSmFeAsOof4.6
B. Superconducting SmFeAsO0.8F0.2 THz (153 cm−1) having similar amplitude at both ~ω .
P
(ii) At the intermediate temperatures above Tc and
The temperature dependence of the ∆R/R-transients low the transientsarepositive onthe sub-ps timescale
F
in the near optimally doped sample is shown in Fig. 4. crosszeroaround4 ps with slowdynamics similar to the
Contrary to the undoped case the transients show com- high-temperatureone. Athigh thepositivepartofthe
F
plextimeandtemperaturedependencies. Independentof transients vanishes and the transients become qualita-
3
ponent has a similar amplitude at both ~ω . At high
P
F
theadditionalnegativecomponentbecomesundetectable
due to a saturation and the transients become virtually
identical to those measuredat 55 K including the coher-
ent phonon response.
Figure 3: Fluence dependence of the ∆R/R transient ampli-
tudeandrelaxation timeinundopedSmAsFeOat twodiffer-
ent temperatures.
Figure 5: Normalized ∆R/R transients at selected temper-
atures at 3.1-eV pump photon energy as a function of F in
superconducting SmFeAsO0.8F0.2. Above ∼150 µJ/cm2 the
tracesstarttooverlapindicatingalinearF-dependence. The
arrows indicate thedirection of increasing F.
II. DISCUSSION
Figure 4: ∆R/R transients as a function of temperature at
A. Undoped SmFeAsO
1.55-eVpump-photonenergy(a)and3.1-eVpump-photonen-
ergy(b)insuperconductingSmFeAsO0.8F0.2. Thepumpflu-
ence was 17 µJ/cm2 at 3.1 eV and 15 µJ/cm2 at 1.55 eV. UponcoolingundopedLaFeAsOundergoesasequence
Below Tc theresponse ofthesuperconductingstateisclearly ofastructuraltransitionfromatetragonaltoorthorhom-
seen. The temperature dependence of the magnetization is bic symmetry, at Ts =156K,and a magnetic SDW tran-
shown in the inset. sition at T = 138K.34 In SmFeAsO T 135K35
SDW SDW
∼
while the structural transition was reported at lower
temperature Ts = 130K.36 So far due to possible differ-
tively identical to those at higher temperatures (see Fig. ent oxygen deficiencies in the two experiments35,36 and
5). The only remainingdifference is a delay-independent strongdopingdependenceofbothTSDWandTsitwasnot
positiveverticalshiftofthetheintermediateT scanswith possible to reliably distinguish between TSDW and Ts in
respectto those measuredat250 K andan increasedco- SmFeAsO.Ourdatashowamarkedcriticalslowingdown
herent phonon frequency of 5.1 THz (170 cm1). The at 125K (see Fig. 1) in the form of the long-lived relax-
~ωP =1.55-eVtransients are,as at higher temperatures, ation tail, while the initial picosecond exponential decay
similar to the ~ωP =3.1-eV transients but weaker. time shows a maximum at 115K. Since the long-lived
∼
(iii) Below T an additional negative component ap- relaxationtailaffectsthequalityofthesingle-exponential
c
pearswitharisetimeof0.2-0.6ps,dependingonthe~ω picosecond fit (see Fig. 2(a)) we can not reliably iden-
P
and pump fluence, and decay time of 5 ps. The com- tify 115 K as a separate transition temperature. We are
∼
4
where Θ is the Debye temperature. From the fit of
D
equation(1)totherelaxationtimeabove230K(seeFig.
2)weobtainλ ω2 =135 10meV2. Ifweestimate ω2
252 meV2 fromh inielastic±neutron data40 we obtaihn λi≈
≈
0.2 indicating a rather weak electron phonon coupling,
which can not explain high T in the doped compound
c
within a single band BCS model. However,owing to the
multibandnatureofiron-pnictidesitispossible,thatdue
toopticalselectionrulessomebandswithpossiblehigher
couplingsarenotdetectedby∆R/Rtransients. Tocheck
the consistencyofthe resultingvalueofλweplotinFig.
2 (b) also the low-T result (2) indicating validity of the
high-T approximation (1) above 230K.
BelowT agapopensattheFermisurfaceintroduc-
SDW
ing a bottleneck in the relaxation. The relaxationacross
a temperature dependent gap was analyzed by Kabanov
etal..16Weuseequation(6)fromKabanovetal.16,which
Figure 6: ∆R/R transients at different pump-photon describes the photo-excited change in quasiparticle den-
energies and selected temperatures in superconducting sity in the presence of a temperature dependent gap, to
SmFeAsO0.8F0.2. For comparison the 295 K transient from fit the amplitude below TSDW = 125 K. Using a single
undoped SmFeAsOis also shown. SDW gapenergy with the BCS temperature dependence
and 2∆ /k T 5 results in a rather good fit to
SDW B SDW
≃
the amplitude temperature dependence (see Fig. 2 (c)).
therefore unable to differentiate between the structural Equation (28) for the relaxation time from Kabanov et
andspintransitionssowewillonlyrefertoasingletran- al.16withthesame∆ (T)describeswellalsothetem-
SDW
sitiontemperatureTSDW Ts 125Kintherestofthe perature dependence of the relaxation time (see Fig. 2
≈ ≈
paper. (c)). However, equation (28) from Kabanov et al.16 also
From our data TSDW is lower than reported in predicts a fast decrease of the relaxation time with ,
literature35,37. The apparent lower TSDW can originate which is not observed in our data. The reason for thFis
in an elevated temperature of the excited volume with mightoriginateinthefactthattheSDWstateisnotfully
respecttothecryostattemperatureduetothelaserheat- gapedandtheenergyrelaxationisnotlimitedbythean-
ing. The 10-K shift of TSDW is larger than the shift harmonic energy transfer from the high frequency to the
∼
of Tc of a few (2-3) K observed in the superconducting low frequency phonons as assumed in the derivation.16
crystals under similar excitation conditions. Due to a
variable thermal coupling to the sapphire substrate for
such small crystals a small decrease of TSDW originating B. Decomposition of the ∆R/R transients in
fromtheoxygendeficiencycannotbedistinguishedfrom superconducting SmFeAsO0.8F0.2 into components.
the laser heating in our samples.
AboveT undopedpnictidesarebadmetalswithre-
SDW
sistivities in the mΩcm range7,38 and plasma frequency
inan 1eVrange6,7. The∆R/Rtransientsinthistem-
∼
perature range are therefore attributed to the relaxation
of electrons in the states near ǫ and can be analyzed
F
by means of the recent theoretical results30 on electron
relaxationinmetals. The -independentrelaxationtime
warrants use of the low exFcitation expansion30, where in
thehightemperaturelimittherelaxationtimeispropor-
tional to the temperature,39
2πk T
B
τ = . (1)
3~λ ω2
h i
Figure 7: Decomposition of ∆R/R transients into different
Here λ ω2 is the second moment of the Eliashberg
functionh30,iλ the electron-phononcoupling constant and components in superconductingSmFeAsO0.8F0.2.
k the Boltzman constant. In the low temperature limit
B
the relaxation time is predicted to diverge at low T:30 While the transients in the undoped sample show a
simplesingle-exponentialrelaxation,whichissensitiveto
2~Θ thephasetransitiontotheorthorhombicSDWstate,the
D
τ = , (2)
π3λk T2 transients in the doped sample show a clear muticompo-
B
5
nentrelaxation. Toseparatecontributionsfromdifferent K, slightly increases with increasing temperature below
components we use41,42 the fluence dependence of the 200 K. Above 200 K, where artifacts due to subtraction
reflectivity transients. The data can be consistently de- of component A start to be significant, τ steeply in-
1B
scribed by three distinct components A, B and C, which creases towards 1 ps. The amplitude of component B,
are tightly connected with three observed temperature A , stays almost constant up to 70 K and then drops
B
∼
regions. monotonously.
The temperature-independent linear scaling of the
transients with above 150 µJ/cm2 suggests the de-
F ∼
composition of the raw ∆R/R into component A which
scales linearly with and a residue which saturates at
F
finite (see Fig. 7). Component A dominates at high
F
temperatures and has to originate in at least three dis-
tinct relaxation processes due to the relatively complex
time evolution. (i) The slower dynamics, which is vir-
tually the same as in undoped SmFeAsO at high tem-
peratures (see Fig. 6), could be attributed to the band
renormalization due to the lattice expansion. (ii) The
sub-ps decay, also having a similar decay time as in un-
doped SmFeAsO at high temperatures, will be discussed
in more detail below. (iii) The oscillatory part of com-
ponent A is attributed to a coherent phonon oscillation
which appears softer as in undoped SmFeAsO. Except
for the shift of the coherent phonon frequency all show
only a minor T-dependence.
The residue shows an unipolar single-exponential de-
cay (see Fig. 8) above T , which we name component
c
B. Component B dominates in the raw transients in the
intermediate temperature range,aboveT . Below T the Figure 8: Temperature dependenceof ∆R/R transients with
c c
residue changestoabipolarmulti-exponentialdecay,ev- component A subtracted. Thin lines are single exponential
identlyduetotheappearanceofanadditionalrelaxation decayfits(componentB)aboveTc andmulti-exponentialde-
processassociatedwiththesuperconductingstatenamed cay fits (asum of component B and C) below Tc.
component C. Component C saturates at lower 10
F ≈
µJ/cm2 than component B, which saturates above 70
µJ/cm2. The decomposition to the three compon∼ents
is further supported by comparison of the raw ∆R/R
at different ~ω shown in Fig. 6 where components A
P
and B show much smaller amplitudes at ~ω = 1.55 eV
P
in comparison to ~ω = 3.1 eV, while the amplitude of
P
component C shows a negligible ~ω dependence.
P
Above T we fit the residue with a single exponential
c
decay (component B )43 (see Fig. 8),
∆RB = ABe−t−τBt0 erfc σ2−4(t−t0)τB , (3)
R 2 (cid:18) 2√2στ (cid:19)
B
where σ correspondsto the effective width of the excita-
tionpulsewithaGaussiantemporalprofilearrivingatt
0
and τ the exponential relaxation time. Below T addi-
B c
tional exponential decays representing component C are
needed to fit the residue,
∆RC =A1C e−tτ−1Ct0 e−tτ−rCt0 Flaixgautrieon9:tiTmeemp(ear)ataunrde daempepnlidtuendcee(obf)thine csoumpperocnoenndtu-Bctirneg-
R (cid:16) − (cid:17) (4) SmAsFeO0.8F0.2 obtained from the fits. The thin line is the
+A2C e−tτ−2tC0 e−tτ−rCt0 fit for the case of a relaxation over a T-independentgap16,41
(cid:16) − (cid:17) with 2∆PG = 120meV. For comparison, the temperature de-
pendence of the relaxation time and the transient amplitude
where τrC represents the rise time and τiC the decay in undoped SmAsFeO is also shown.
times. The resulting fit parametersfor component B are
shown in Fig. 9. The decay time, τ 0.25 ps at 4
1B
≈
6
C. Superconducting response in SmFeAsO0.8F0.2
For easier separation of component C associated with
the superconducting response we use the fact that com-
ponents A and B are temperature independent in the
superconducting state. We therefore extract component
C by subtracting the average of transients measured at
55 and 65 K from transients measured below 55 K. The
subtractedtransientsclearly showa two-stepdecaywith
a finite rise time and are excellently fit by equation (4)
as shown in Fig. 10.
Figure 11: Rise time and relaxation time (a), (c) and ampli-
tude (b), (d) of component C as functions of temperature at
F =2.9µJ/cm2 and3µJ/cm2 for~ωP =1.55eVand3.1eV,
respectively,(a),(b)andatF =17.4µJ/cm2 and15µJ/cm2
for ~ωP =1.55 eVand3.1 eV,respectively (c),(d). Forcom-
Figure 10: Component C as a function of temperature at
parison amplitudes in a slightly higher Tc sample are shown
different pump photon energies and fluences. Thin lines are
in (b), (d). The thin line in (c) is the fit of equation (5) to
fitsdiscussed in text.
thedata.
In the spirit of the Rotwarf-Taylor model27 we asso-
ciate the rise time with establishment of the thermal
quasi-equilibrium between the photo-excited quasiparti-
ternal fluence, , at which the superconductivity is
cles and high frequency (~ω > 2∆ ) phonons. The FT
SC destroyed in the most excited spot of the pump beam.
shorter relaxation time is associated with establishment
From we calculate, using optical penetration depths,
T
of the local thermal equilibrium between all degrees of λop, aFnd reflectivities, R, of LaAsFeO1−xFx,44 the en-
freedom, while the longer relaxation time is due to en-
ergy density, U , required to completely destroy the su-
p
ergyescape outofthe probedvolume. This is supported
breyspinecctretaosetdheretolattailveamapmliptluitduedaetohfigthheerlFon.g decay with p((UUerppco==nd118u2cJtJ/i/nmmgoosll)t)aatatet:~~ωUωpP/Pk=B=3=.11.F5eT5V(1e−aVnR.d)/TλUhoppe/kkBaBv==era21g..25e KKva//lFFueee
1ta1lN)eserhirtoohwre,rawnthhyieltereoimsneplyteirmτateunsrheoordwtehpseednerdepleeannxcdaeetniwocinethtoiinnmeexsap(neserdiem~Fωeing-.. sUepr/vkBed=in1.L8aK1−/xFSerxisCsulOig4h.t3l1yIsfmwaellearsstuhmane tthheatvathlueesthoebr--
τ is always faster arCt ~ω = 1.55 eV and sFhows mucPh modynamicsuperconductingcondensationenergy,Uc,in
wcrreCeaakseersFfro-dmep0e.3ndpesnacte tha=nP3atµJ~/ωcPm=2 t3o.10e.6Vp,swahterei=t i1n5- SlamrFmeaAgsnOit0u.8dFe0s.2ofisTscimweilaorbttoaiLnaU1p−/xUScr≫xC1uOin4didcuaetitnogstimhait-
µJ/cm2. While an incrFease of τ with increasingF~ω is a significant amount of excitation energy is transferred
rC P to the bath on a timescale of 0.3 ps. If all degrees of
expected due to the photo-excitation being farther away ∼
freedom would absorb U the resulting temperature rise
fromǫ anincreasing -dependencewithincreasing~ω p
F F P would be 11K based on the published heat capacity, cp,
is not completely understood. data45. Contraryto the cuprates c is dominated by Sm
p
In Fig. 12 we plot -dependence of the component spins45,46 below 12Ksoitisnotpossibletodetermine
C amplitude, A . SiFmilarly as in the cuprates31 the ∼
SC whether the excess U is absorbed in the Sm-spin or in
p
response saturates with increasing indicating a com-
the phonon subsystem.
F
plete destruction of the superconducting state in the ex-
cited volume. By taking into account the effects of in- We fit the temperature dependence of the reflectivity
homogeneous excitation due to finite penetration depths change upon complete destruction of the superconduct-
and beam diameters31 we determine the threshold ex- ingstateshowninFig. 11(d)bythehigh-frequencylimit
7
of the Mattis-Bardeen formula,47 channelscorrespondtotwodistinctelectronicsubsystems
which are weakly coupled on the sub-ps timescale. One
∆R ∆(T) 2 3.3~ω exhibitsthesuperconductinggap(s)andtheotherapseu-
log , (5)
R ∝(cid:18) ~ω (cid:19) (cid:18)∆(T)(cid:19) dogap.
A possible origin of the distinct electronic subsystems
where ~ω is the probe-photon energy and ∆(T) the su- could be a chemical phase separation of the doped F.
perconducting gap. By using the BCS-gap temperature However,thesuperconductingtransitionisrathernarrow
dependence with 2∆0/kBTc = 3.5 we obtain an excellent (see Fig. 4) so the presence of weakly-superconducting
fit to the observed temperature dependence. Unfortu- fluorine-poor regions in which SDW is suppressed giv-
nately the shape of the temperature dependence (5) is a ingrisetoadditionalpseudo-gappedelectronicsubsystem
very weak function of 2∆0/kBTc and one can not reliably is also unlikely. More importantly undoped SmFeAsO
distinguish the contributions from different gaps48 and showsvirtuallynopump-photonenergydispersionwhich
reliably determine 2∆0/kBTc. the superconducting sample does, so a simple chemical
phase separation to doped and undoped regions is very
unlikely despite similarity (see Fig. 6) between compo-
nent A and the undoped SmFeAsO room-temperature
transients. Moreover, there is no divergent signal at 125
K (or anywhere near that temperature) which can be
attributed to the presence of the undoped phase in the
superconducting sample. We therefore believe that both
electronic subsystems are intrinsic to the SC material
Apart from the chemical phase separation an intrin-
sic electronic phase separationakin to that proposed for
the cuprates49 could be origin of the distinct electronic
subsystems. The existence of intrinsic electronic phase
separation has been reported in 122 systems,50,51 how-
ever at present the issue is still rather controversial.
In the case of the spatially homogeneous electronic
Figure12: TheamplitudeofthecomponentCasfunctionsof state different electronic subsystems would correspond
fluenceat different ~ωP. Thin lines are fitsdiscussed in text. todifferentbandscrossingǫ .Thiswouldimply thatthe
F
inter-band scattering between parts of the Fermi surface
correspondingto different electronic subsystems is negli-
gible on a timescale of a few hundred fs, since excitation
D. Normal state response in SmFeAsO0.8F0.2 photon at 1.5 eV only weakly excites components A and
B.Further,sincebothcomponentsexistinthesupercon-
The temperaturedependence ofthe component-Bam- ductingstateevenatlow thepartoftheFermisurface
F
plitude is consistent with a bottleneck due to the relax- correspondingtocomponentsAandBhastoremainun-
ation over a T-independent gap16 (see Fig. 9) with a gapped or pseudo-gaped in the superconducting state.
magnitude 2∆ = 120 meV as noted previously.41 The Anotherpossibilityforaweaklycoupledelectronicsub-
PG
sub-ps part of component A on the other hand is tem- system are Sm crystal-field levels. The energy of the
perature independent suggesting finite density of states levels in SmFeAsO1−xFx is in the range of 20-60 meV
at ǫ . This is consistent with heat capacity measure- as determined indirectly from heat capacity fits.52,53 In-
F
ments in polycrystalline SmFeAsO1−xFx45,46 where a fi- volvement of the crystal-field levels could explain the
nite Sommerfeld constant in the superconducting state strong ~ωP-dependence of components A and B and de-
suggestsafinite densityofstatesatǫ . Due to verysim- couplingfromtheotherlowlyingelectronicstates. How-
F
ilarpump-photonenergydispersionofcomponentsAand ever, the observation of a pseudogap by NMR in La-
Bwebelievethattheyoriginatefromthesameelectronic 111112,13 (which has no crystal field level structure at
states which have a soft-gapped density of states at ǫF. low energy), and the weak ~ωP-dependence in undoped
Saturationof component B amplitude with increasing SmFeAsO point against such a scenario.
F
indicatesthatthepseudogapcanbedestroyedsoitisnot Finally, let us briefly compare our results in Sm-1111
a simple band-structure effect. tofemtosecondspectroscopyin(Ba,K)-122.54,55 Thereis
The pump-photon energy dispersion similar to that a marked difference in the magnitude of the relaxation
of components A and B is not observed for component time in the superconducting state, which increases with
C. Electronic states involved in the relaxationrelated to decreasingtemperaturebeyond60psinoptimallydoped
componentsAandBmustthereforebedifferentthanfor (Ba,K)-12254 and remains T-independent in Sm-1111 at
componentC.ThisconfirmsthattherelaxationbelowT 5 ps. Similarly, the excitation fluence dependence of
c
does not proceed via a cascade but rather through dis- ∼the relaxation time, which is absent in Sm-1111 is pro-
tinct parallel channels as suggested previously.41 These nounced in (Ba,K)-122.55 This suggests different relax-
8
ation mechanisms in Sm-1111 and (Ba,K)-122. While of component B suggests the presence of a temperature
behavior in (Ba,K)-122 is consistent with the Rotwarf- independent pseudogap with a magnitude 2∆ 120
PG
Taylor model27,56 where the anharmonic optical-phonon meV. The pseudogap is destroyed at a finite fluen≃ce in-
decay is determining the relaxation time in Sm-1111 the dicating that it is not a band-structure effect (such as
presence of the ungapped electronic subsystem seems to a 120 meV gap at some arbitrary point in the Bril-
provide a competing relaxation channel. However, only louin zone). Component C is observed only in the su-
~ω = 1.55 eV was used in (Ba,K)Fe As so ultrafast perconducting state and corresponds to the relaxation
P 2 2
pump-probe spectroscopy at different ~ω in 122 sys- across a T-dependent superconducting gap with a BCS
P
tems and measurements in LaFeAsO1−xFx are needed temperature dependence. At high enough pump flu-
for determination whether some fundamental difference enceacompletedestructionofthesuperconductingstate
between 1111 and 122 systems is responsible for the dif- is observed with the critical optical excitation density
ferentrelaxationtime behaviorandmarkedtemperature Up/k 1.8 K/Fe which is similar to the value observed
B ≈
dependence above T in Sm-1111. in (La,Sr)CuO .
c 4
ThemulticomponentrelaxationinSCsamplesstrongly
suggest the presence of two relatively weakly coupled
III. SUMMARY AND CONCLUSIONS electronic subsystems, one exhibiting the SC gap(s)
and the other the pseudogap. From the temperature
In undoped SmFeAsO a single-exponential relaxation and fluence dependence of photoinduced optical reflec-
is observed. Fromthe high-T relaxationtime the second tivity transients in undoped and near-optimally doped
moment of the Eliashaberg function is determined to be SmFeAsO1−xFx single crystals it is clear that the pres-
λ ω2 =135 10meV2. The couplingconstantλ 0.2, enceoftwoelectronicsubsystemsinthesuperconducting
± ≈
es(cid:10)tim(cid:11)atedfromthisvalue,iscomparabletolow-Tcsuper- sample is not a result of a simple phase separation. The
conductorsandcannotexplainthe highsuperconducting fact that no relaxation component - such as appears in
TcsofthesecompoundswithinasinglebandBCSmodel. the SDW phase - is seen in the SC phase appears to
BelowT the temperaturedependence ofthe relax- rule this out. The presence of two electronic subsystems
SDW
ationindicatesappearanceofaQPrelaxationbottleneck thereforeoriginateseitherinanintrinsicphaseseparation
due to opening of a single charge gap at T with a or more likely in the multiband nature of the supercon-
SDW
BCS-like temperature dependence and the amplitude of ducting iron-pnictides.
2∆ /k T 5 at 4.2 K. A question whether this
SDW B SDW
≃
chargegapisadirectconsequenceoftheSDWformation
or due to the structural transition unfortunately cannot
Acknowledgments
be answered from our data.
In superconducting SmFeAsO0.8F0.2 three distinct re-
laxation components are observed. Components A and ThisworkhasbeensupportedbyARRS(GrantNo.P1-
B are present in both the superconducting and the nor- 0040)andthe SwissNationalScience FoundationNCCR
malstate. Thetemperaturedependenceoftheamplitude MaNEP.
1 Y. Kamihara, H. Hiramatsu, M. Hirano, R. Kawamura, Arxiv preprintarXiv:0910.1318 (2009).
H. Yanagi, T. Kamiya, and H. Hosono, Journal of the 8 H. Ding, P. Richard, K. Nakayama, K. Sugawara,
American Chemical Society 128, 10012 (2006), URL T. Arakane, Y. Sekiba, A. Takayama, S. Souma, T. Sato,
http://dx.doi.org/10.1021/ja063355c. T.Takahashi,etal., EPL(EurophysicsLetters)83,47001
2 Y.Kamihara, T.Watanabe,M.Hirano, H.Hosono, etal., (2008).
J. Am.Chem. Soc 130, 3296 (2008). 9 S. Kawasaki, K. Shimada, G. F. Chen, J. L.
3 Z.Ren,G.Che,X.Dong,J.Yang,W.Lu,W.Yi,X.Shen, Luo, N. L. Wang, and G. qing Zheng, Physi-
Z.Li,L.Sun,F.Zhou,etal.,EPL-EurophysicsLetters83, cal Review B (Condensed Matter and Materi-
17002 (2008). als Physics) 78, 220506 (pages 4) (2008), URL
4 D. J. Singh and M.-H. Du, Physical Review http://link.aps.org/abstract/PRB/v78/e220506.
Letters 100, 237003 (pages 4) (2008), URL 10 D. Daghero, M. Tortello, R. S. Gonnelli, V. A.
http://link.aps.org/abstract/PRL/v100/e237003. Stepanov, N. D. Zhigadlo, and J. Karpinski, Phys-
5 A. I. Coldea, J. D. Fletcher, A. Carrington, J. G. An- ical Review B (Condensed Matter and Materi-
alytis, A. F. Bangura, J.-H. Chu, A. S. Erickson, I. R. als Physics) 80, 060502 (pages 4) (2009), URL
Fisher, N. E. Hussey, and R. D. McDonald, Physi- http://link.aps.org/abstract/PRB/v80/e060502.
cal Review Letters 101, 216402 (pages 4) (2008), URL 11 C. Martin, M. E. Tillman, H. Kim, M. A. Tanatar,
http://link.aps.org/abstract/PRL/v101/e216402. S. K. Kim, A. Kreyssig, R. T. Gordon, M. D. Van-
6 W. Hu, J. Dong, G. Li, Z. Li, P. Zheng, G. Chen, J. Luo, nette, S. Nandi, V. G. Kogan, et al., Physical Re-
and N.Wang, PhysRev Lett 101, 257005 (2008). view Letters 102, 247002 (pages 4) (2009), URL
7 Z.Chen,R.Yuan,T.Dong,N.Wang,S.Ce,N.Pr,etal., http://link.aps.org/abstract/PRL/v102/e247002.
9
12 K. Ahilan, F. L. Ning, T. Imai, A. S. Sefat, 34 H.-H. Klauss, H. Luetkens, R. Klingeler, C. Hess, F. J.
R. Jin, M. A. McGuire, B. C. Sales, and D. Man- Litterst, M. Kraken, M. M. Korshunov, I. Eremin,
drus, Physical Review B (Condensed Matter and Ma- S.-L. Drechsler, R. Khasanov, et al., Physical Re-
terials Physics) 78, 100501 (pages 4) (2008), URL view Letters 101, 077005 (pages 4) (2008), URL
http://link.aps.org/abstract/PRB/v78/e100501. http://link.aps.org/abstract/PRL/v101/e077005.
13 Y.Nakai,S.Kitagawa,K.Ishida,Y.Kamihara,M.Hirano, 35 A.Drew, C.Niedermayer,P.Baker,F.Pratt, S.Blundell,
andH.Hosono,NewJournalofPhysics11,045004(2009). T.Lancaster,R.Liu,G.Wu,X.Chen,I.Watanabe,etal.,
14 R. Gonnelli, D. Daghero, M. Tortello, G. Um- Naturematerials (2009).
marino, V. Stepanov, R. Kremer, J. Kim, 36 S. Margadonna, Y. Takabayashi, M. McDonald,
N. Zhigadlo, and J. Karpinski, Physica C: Su- M. Brunelli, G. Wu, R. Liu, X. Chen, and K. Pras-
perconductivity 469, 512 (2009), ISSN 0921- sides, Physical Review B 79, 14503 (2009).
4534, superconductivity in Iron-Pnictides, URL 37 S. Sanna, R. De Renzi, G. Lamura, C. Ferdeghini,
http://www.sciencedirect.com/science/article/B6TVJ-4VWB1A9T.-P8/a2le/n5z6o0n2a3,afM7.edP6u0t7t2i,85Mf0.cTbbro7pce6afnco0,35adncd4.T. Shiroka,
15 C.J.Stevens,D.Smith,C.Chen,J.F.Ryan,B.Podobnik, Physical Review B 80, 52503 (2009).
D.Mihailovic,G.A.Wagner,andJ.E.Evetts,Phys.Rev. 38 H.Luo,Z.Wang,H.Yang,P.Cheng,X.Zhu,andH.Wen,
Lett. 78, 2212 (1997). Supercond.Sci. Technol 21, 125014 (2008).
16 V. V. Kabanov, J. Demsar, B. Podobnik, and D. Mi- 39 C. Gadermaier, A. S. Alexandrov, V. V. Kabanov,
hailovic, Phys. Rev.B 59, 1497 (1999). P. Kusar, T. Mertelj, X. Yao, C. Manzoni, G. Cerullo,
17 J. Demsar, B. Podobnik, V. V. Kabanov, T. Wolf, and and D. Mihailovic, ArXiv e-prints(2009), 0902.1636.
D. Mihailovic, Phys. Rev.Lett. 82, 4918 (1999). 40 R.Osborn,S.Rosenkranz,E.Goremychkin,andA.Chris-
18 R. Kaindl, M. Woerner, T. Elsaesser, D. Smith, J. Ryan, tianson,PhysicaC:Superconductivityanditsapplications
G. Farnan, M. McCurry, and D. Walmsley, Science 287, 469, 498 (2009).
470 (2000). 41 T. Mertelj, V. Kabanov, C. Gadermaier, N. Zhigadlo,
19 D. Dvorsek, V. V. Kabanov, J. Demsar, S. M. Kazakov, S. Katrych, J. Karpinski, and D. Mihailovic, Physical Re-
J. Karpinski,and D.Mihailovic, Phys. Rev.B 66, 020510 view Letters 102, 117002 (2009).
(2002). 42 T. Mertelj, V. Kabanov, C. Gadermaier, N. Zhigadlo,
20 G.P.Segre,N.Gedik,J.Orenstein,D.A.Bonn,R.Liang, S.Katrych,Z.Bukowski,J.Karpinski,andD.Mihailovic,
and W.N. Hardy,Phys.Rev.Lett. 88, 137001 (2002). Journal of Superconductivity and Novel Magnetism 22,
21 M.Schneider,J.Demsar,Y.Glinka,A.Klimov,A.Krapf, 575 (2009).
S.Rast,Y.Ren,etal.,EPL(EurophysicsLetters)60,460 43 D. Mihailovic and J. Demsar, Spectroscopy of Supercon-
(2002). ducting Materials (American Chemical Society: Washing-
22 J. Demsar, R. D. Averitt, A. J. Taylor, V. V. Kabanov, ton, DC, 1999), vol. 730 of ACS Symposium Series, chap.
W. N. Kang, H. J. Kim, E. M. Choi, and S. I. Lee, Phys. Time-resolved optical studiesof quasiparticle dynamicsin
Rev.Lett. 91, 267002 (2003). high-temperaturesuperconductors, pp.230–244.
23 N.Gedik,J.Orenstein,R.Liang,D.Bonn,andW.Hardy, 44 A. V. Boris, N. N. Kovaleva, S. S. A. Seo,
Science 300, 1410 (2003). J. S. Kim, P. Popovich, Y. Matiks, R. K. Kre-
24 N. Gedik, P. Blake, R. C. Spitzer, J. Orenstein, R. Liang, mer, and B. Keimer, Physical Review Let-
D. A. Bonn, and W. N. Hardy, Phys. Rev. B 70, 014504 ters 102, 027001 (pages 4) (2009), URL
(2004). http://link.aps.org/abstract/PRL/v102/e027001.
25 P. Kusar, J. Demsar, D. Mihailovic, and S. Sugai, Phys. 45 M. Tropeano, A. Martinelli, A. Palenzona, E. Bellingeri,
Rev.B 72, 014544 (2005). E. G. d’Agliano, T. D. Nguyen, M. Affronte, and
26 R. A. Kaindl, M. A. Carnahan, D. S. Chemla, S. Oh, and M. Putti, Physical Review B (Condensed Matter and
J. N.Eckstein, Phys. Rev.B 72, 060510 (2005). Materials Physics) 78, 094518 (pages 7) (2008), URL
27 V.V.Kabanov,J.Demsar,andD.Mihailovic, Phys.Rev. http://link.aps.org/abstract/PRB/v78/e094518.
Lett. 95, 147002 (2005). 46 L. Ding, C. He, J. K. Dong, T. Wu, R. H. Liu, X. H.
28 G. Bianchi, C. Chen, M. Nohara, H. Takagi, and J. F. Chen,andS.Y.Li,PhysicalReviewB(CondensedMatter
Ryan,Phys. Rev.B 72, 094516 (2005). and Materials Physics) 77, 180510 (pages 4) (2008), URL
29 M.L.Schneider,M.Onellion,X.X.Xi,X.Zeng,A.Souki- http://link.aps.org/abstract/PRB/v77/e180510.
assian, P.Omernik,andG.Taft,Phys.Rev.B70,012504 47 D.C.Mattis andJ.Bardeen,Phys.Rev.111, 412 (1958).
(2004). 48 J. Karpinski, N. Zhigadlo, S. Katrych, Z. Bukowski,
30 V. V. Kabanov and A. S. Alexandrov, Physi- P. Moll, S. Weyeneth, H. Keller, R. Puzniak,
cal Review B (Condensed Matter and Materi- M. Tortello, D. Daghero, et al., Physica C: Su-
als Physics) 78, 174514 (pages 8) (2008), URL perconductivity 469, 370 (2009), ISSN 0921-
http://link.aps.org/abstract/PRB/v78/e174514. 4534, superconductivity in Iron-Pnictides, URL
31 P. Kusar, V. Kabanov, J. Demsar, T. Mertelj, S. Sugai, http://www.sciencedirect.com/science/article/B6TVJ-4VWB19T-4/
and D. Mihailovic, Physical Review Letters 101, 227001 49 L. Gor’kov, Journal of Superconductivity 14, 365 (2001).
(2008). 50 J. T. Park, D. S. Inosov, C. Niedermayer, G. L. Sun,
32 N. Zhigadlo, S. Katrych, Z. Bukowski, S. Weyeneth, D. Haug, N. B. Christensen, R. Dinnebier, A. V.
R. Puzniak, and J. Karpinski, J. Phys.: Condens. Mat- Boris, A. J. Drew, L. Schulz, et al., Physical Re-
ter 20, 342202 (2008). view Letters 102, 117006 (pages 4) (2009), URL
33 V. G. Hadjiev, M. N. Iliev, K. Sasmal, Y.-Y. Sun, and http://link.aps.org/abstract/PRL/v102/e117006.
C. W. Chu, Physical Review B (Condensed Matter and 51 Y. Laplace, J. Bobroff, F. Rullier-Albenque, D. Colson,
Materials Physics) 77, 220505 (pages 3) (2008), URL and A. Forget, Physical Review B (Condensed Matter
http://link.aps.org/abstract/PRB/v77/e220505. and Materials Physics) 80, 140501 (pages 4) (2009), URL
10
http://link.aps.org/abstract/PRB/v80/e140501. 55 D.H.Torchinsky,G.F.Chen,J.L.Luo,N.L.Wang,and
52 R. Cimberle, C. Ferdeghini, F. Canepa, M. Ferretti, N.Gedik, ArXiv e-prints(2009), 0905.0678.
A. Martinelli, A. Palenzona, A. Siri, and M. Tropeano, 56 A. Rothwarf and B. N. Taylor, Phys. Rev. Lett. 19, 27
Arxiv preprint arXiv:0807.1688 (2008). (1967).
53 P. Baker, S. Giblin, F. Pratt, R. Liu, G. Wu, X. Chen, 57 K. Ishida, Y. Nakai, H. Hosono, Y. Kuramoto,
M. Pitcher, D. Parker, S. Clarke, and S. Blundell, New H. Kusunose, A. Kiss, R. Kubo, M. Fujita, K. Wak-
Journal of Physics 11, 5010 (2009). abayashi, K. Nakada, et al., J. Phys. Soc. Jpn 78, 062001
54 E. E.M. Chia, D.Talbayev,J.Zhu,H.Q.Yuan,T. Park, (2009).
J. D. Thompson, G. F. Chen, J. L. Luo, N. L. Wang, and 58 For a recent review see ref. 57.
A. J. Taylor, ArXive-prints(2008), 0809.4097.