Temperature dependence of the non-local spin Seebeck effect in YIG/Pt nanostructures

4.9 MB·

by  Kathrin Ganzhorn

Temperature dependence of the non-local spin Seebeck effect in YIG/Pt
nanostructures
Kathrin Ganzhorn,1,2,a) Tobias Wimmer,1,2 Joel Cramer,3,4 Richard Schlitz,1,2,5 Stephan Gepr¨ags,1 Gerhard
Jakob,3 Rudolf Gross,1,2,6 Hans Huebl,1,2,6 Mathias Kl¨aui,3,4 and Sebastian T.B. Goennenwein1,2,5,6
1)Walther-Meißner-Institut, Bayerische Akademie der Wissenschaften, 85748 Garching,
Germany
2)Physik-Department, Technische Universita¨t Mu¨nchen, 85748 Garching, Germany
3)Institute of Physics, Johannes Gutenberg-University Mainz, 55099 Mainz, Germany
4)Graduate School of Excellence Materials Science in Mainz, Staudinger Weg 9, 55128 Mainz,
Germany
7 5)Institut fu¨r Festko¨rperphysik, Technische Universit¨at Dresden, 01062 Dresden,
1 Germany
0 6)Nanosystems Initiative Munich, 80799 Munich, Germany
2
We study the transport of thermally excited non-equilibrium magnons through the ferrimagnetic insulator

n YIG using two electrically isolated Pt strips as injector and detector. The diffusing magnons induce a non-
a
local inverse spin Hall voltage in the detector corresponding to the so-called non-local spin Seebeck effect
J
(SSE). We measure the non-local SSE as a function of temperature and strip separation. In experiments at
0
room temperature we observe a sign change of the non-local SSE voltage at a characteristic strip separation
1
d , in agreement with previous investigations. At lower temperatures however, we find a strong temperature
0
] dependence of d0. This suggests that both the angular momentum transfer across the YIG/Pt interface as
l well as the transport mechanism of the magnons in YIG as a function of temperature must be taken into
l
a account to describe the non-local spin Seebeck effect.
h
-
s
e Magnons,thecollectiveexcitationsinmagneticallyor- signal is not well established. In this context, the non-
m
dered systems, represent an attractive option for infor- localSSEhasrecentlybeenstudiedatroomtemperature
. mation transfer and processing. Using the ferrimagnetic as a function of strip separation d and YIG thickness11.
t
a insulator Yttrium Iron Garnet (YIG) as a model sys- Whileatshortdistancesdthelocalandthenon-localSSE
m tem,magnon-basedinformationprocessingschemeshave signals have the same sign, for larger distances the non-
- been put forward on the basis of coherently excited spin local SSE amplitude is inverted, which was attributed to
d waves1–3. Recent experiments in YIG/Pt heterostruc- theprofileofthenon-equilibriummagnondistributionin
n
turesfurthermoreshowthatinformationcanalsobecar- the YIG film.
o
c ried by incoherent non-equilibrium magnons4,5 diffusing In this letter, we systematically study the non-local
[ in YIG. This approach even allows for the implementa- spin Seebeck effect in YIG/Pt nanostructures as a func-
tionoflogicoperationswithinthemagneticsystem6. The tion of temperature and strip separation and find that
1
non-equilibrium magnons can be excited and detected thenon-localSSEvoltagechangessignatacharacteristic
v
5 electricallyviaspinscatteringmechanismsattheYIG/Pt stripseparationd0,whichisstronglytemperaturedepen-
3 interface. In a scheme referred to as magnon-mediated dent. We interpret our findings as evidence of a complex
6 magnetoresistance (MMR), magnons are generated by interplay between the temperature dependences of the
2 driving a dc charge current through an injector Pt strip interfacialtransparency,i.e. angularmomentumtransfer
0 and detected as a non-local voltage in a second strip. across the YIG/Pt interface, and the diffusive properties
.
1 The MMR effect has been studied as a function of the of the thermally excited non-equilibrium magnons.
0 distance d between injector and detector4, temperature5
We investigate the non-local spin Seebeck effect in
7 and magnetic field magnitude and orientation7, allow-
YIG/Pt bilayers fabricated and nano-patterned at the
1 ing for the extraction of the length scales involved in the
: Walther-Meißner-Institut (sample series A) and at Jo-
v magnondiffusionprocess. Inadditiontotheelectricalin-
hannes Gutenberg-University Mainz (sample series B).
i jection, non-equilibrium magnons can also be generated
X Series A was fabricated starting from a commercially
thermally, via local Joule heating in the injector strip. available 2µm thick YIG film grown onto (111) oriented
r
Theensuingthermalnon-localvoltageiscallednon-local
a Gd Ga O (GGG) via liquid phase epitaxy (LPE). Af-
3 5 12
spin Seebeck effect in analogy to the well established lo-
ter Piranha cleaning and annealing (see Ref. 12 for de-
cal (longitudinal) spin Seebeck effect (SSE)8. While the
tails) to improve the interface quality, 10nm of Pt were
microscopic mechanisms and in particular the relevant
deposited onto the YIG film using electron beam evapo-
length scales for the SSE have been investigated in quite
ration. Aseriesofnanostructuresconsistingof2parallel
somedetail9,10,thephysicsbehindthenon-localthermal Pt strips with length l=100µm, width w =500nm and
an edge-to-edge separation of 20nm ≤ d ≤ 10µm were
patterned using electron beam lithography followed by
Ar ion etching. Series B was fabricated using a 3.3µm
a)Electronicmail: [email protected] thickLPE-YIGfilmgrownontoaGGGsubstrateaswell.
2
J + - 300K). For series A an external magnetic field µ0H =
c
(a) 1T was rotated in the thin film plane, while for series
V B the external magnetic field was applied along the y-
+ -loc directionandsweptfrom−250mTto+250mT(α=90◦,
270◦ in Fig. 1 (b)). For local longitudinal spin Seebeck
effectmeasurementsinonesingle(injector)strip,weused
10µm
thecurrentheatingmethoddescribedinRef.13: acharge
current J = 100µA is applied to the Pt strip along the
c
x direction using a Keithley 2400 source meter, inducing
Joule heating in the normal metal. The ensuing temper-
V
ature gradient across the Pt/YIG interface gives rise to
nl
+ - the spin Seebeck effect and generates a spin current J
s
flowing across the interface, with the spin current spin
H polarization s determined by the orientation of the mag-
(b) netization M in YIG. This spin current is accompanied
α
Hi z by a charge current JISHE flowing along the x-direction
Pt in the Pt, as shown in Fig. 1 (b). The voltage drop Vloc,
Lo JISHIEnjeLcotorDetector 1 HiPt y Hi x wtvghoihnelti,cmitnhhejieetncecStrlo.SurESdsietnvsrociteplhtateihsgleeorecsicsapoliprnSdrSoSepEdeeouabrntseiidconkngtaheaffleetKroceetstiitihsshtelioevJfyeotuhr2el1eesr8pm2hoeannalsateoninrooig--f
s J s JISHE Detector 2JISHE Pt pcwuoerwreeexrnttirnadctirtheecVtPiotna.nUdstihn=egret(fhVoeres(wi+nitdJcehp)ien+ngdVsecnhte(mo−feJtho)ef)/Rh2ee,af.tai1nn3dg,
s J Lo thereby elimitnhaetrme,alodcditionalolcresisctive effloeccts scuch as the
s spin Hall magnetoresistance.
s Figure 2 (a) shows V measured as a function
therm,loc
G
J YI of the magnetic field orientation α with respect to the
s x-axis at 50K for a device from series A. We observe
the characteristic SSE dependence V ∝ sin(α)
therm,loc
yieldingapositiveamplitudeA =V (90◦)−
SSE,loc therm,loc
V (270◦) of the local SSE, as expected in YIG/Pt
therm,loc
heterostructures for this field configuration14.
FIG. 1. (a) Optical micrograph of a non-local nanostructure Using an additional nanovoltmeter, we simultane-
withtwoPtstrips(bright)onaYIGfilm(dark),includingthe aously measure the voltage drop V arising along the
nl
electrical wiring. (b) Sketch of the YIG/Pt heterostructure:
unbiased and electrically isolated second Pt strip. In
a dc charge current J (not shown here) is applied to the
c analogy to the local thermal signal, the non-local ther-
injector strip (left) and the spin Seebeck signal is detected
mal voltage is extracted as V = (V (+J ) +
locally via the ISHE. The corresponding non-local thermal therm,nl nl c
V (−J ))/2 in order to distinguish it from resis-
signalismeasuredalongthedetectorstripsfordifferentstrip nl c
separations. The color coding gives a qualitative profile of tive non-local effects such as the magnon mediated
the magnon accumulation µm in the YIG film, where red magnetoresistance5. Vtherm,nl as a function of the ex-
corresponds to µ < 0 (magnon depletion) and blue to a ternal magnetic field orientation at 50K is depicted in
m
positive µ (magnon accumulation). In the short distance Fig. 2 (b) and (c) for strip separations of d = 200nm
m
regime µm < 0 at the injector and detector, such that the and2µm(seriesA).Inbothdevices, weobserveasin(α)
same sign is expected for local and non-local SSE. With in- dependence, with an amplitude A about one order
SSE,nl
creasingdistancefromtheinjector,µ andconsequentlythe
m of magnitude smaller than for the local SSE. While the
spincurrentacrosstheinterfaceaswellasthedetectedISHE
signal is positive for the d = 200nm device, a negative
voltage change sign.
A is observed in the device with a larger injector-
SSE,nl
detector separation of d=2µm. In order to confirm this
sign change, we extract the amplitude of the non-local
A series of strips with width w =1µm and edge-to-edge SSEmeasuredat50Kindifferentdeviceswithstripsep-
distance 100nm≤d≤1.2µm were patterned using elec- arationsrangingfrom20nmto10µm. Theresultingdata
tron beam lithography followed by a lift-off process with is shown in Fig. 2 (d) as green symbols. Indeed, a sign
a Pt thickness of 7.5nm. An optical micrograph of one change is observed at a strip separation d ≈ 560nm.
0
of these nanostructures is depicted in Fig. 1 (a). Repeating these measurements as a function of temper-
In order to study the devices in series A and B, the ature in the range between 10K and 300K yields the
samples were mounted in the variable temperature in- data compiled in Fig. 2 (d). For all temperatures, a
sert of a superconducting magnet cryostat (10K ≤ T ≤ sign change in A is observed as a function of the
SSE,nl
3
(a) d=0 (local) (b) d=0.2µm (c) d=2µm
20 0.5 0.5 4
µV) 50K µV) 50K µV) 4000
20 V(therm,loc -2000°ASSE, l1oc800°.5360° V(therm,nl-0.050°ASSE, n1l080°.5360°V(therm,nl-0.050°ASSE, n1l803°5000K3060° m) 3
α α α µ 2
0 0.0 0.0 2000 d (0
(d) 03 180360 0 180360 0 180360
-320 -0.5 -0.5 323200000000KKKK 1000 1 SSSSeeaarrmmiieeppssllee ABAB
2
2 A(µV)SSE, nl 1 15311531000500050KKK0KKKKK 0 0 100 T (K) 200 300
1
1100KK
0 0 100 200 300
0 FIG. 3. Temperature dependence of the critical strip separa-
tion d at which the non-local SSE changes sign for sample
0
-1
series A (red) and B (blue).
0 2 4 6 8 10
-1
d(µm)

FIG. 2. (a) Local spin Seebeck voltage detected at the injec- al.11, this leads to a depletion of magnons (µ < 0, red
0 2 4 6 8 10 m
tor strip at 50K as a function of the in-plane magnetic field inFig.1(b))comparedtothethermalequilibriumpopu-
orientation α with respect to the x axis. (b), (c) Non-local lationbeneaththeinjector. Ontheotherhand, diffusing
thermal voltage detected at T = 50K at the second strip
magnonsaccumulatefurtherawayfromtheinjector,giv-
for a strip separation of d = 200nm and 2µm, respectively.
ing rise to µ > 0 (blue in Fig. 1 (b)). This model
(d)Non-localSSEamplitudeA extractedfromin-plane m
SSE,nl is applied to describe the increase of d with increasing
field rotations at temperatures between 10K and 300K as a 0
YIG thickness observed by Shan et al.11: in contrast to
function of the injector-detector separation d.
phonons,themagnonscannotcrosstheYIG/GGGinter-
face and accumulate there. As a consequence, d (which
0
marks the sign change of µ ) shifts to smaller values
m
strip separation. Invariably, for small gaps the local and
for thinner YIG films. Note that since the overall pro-
non-local SSE are both positive, but for large gaps the
file of µ is governed by diffusive magnon transport, the
m
non-local SSE becomes negative. The experimental data corresponding length scales can reach several µm11. As
in Fig. 2 (d) show that the critical strip separation d ,
0 shown in Fig. 1 (b), the sign of µ determines the di-
m
which is defined by A = 0, shifts to larger values as
SSE rection of the interfacial spin current J at the detector,
s
the temperature increases. The values d extracted from
0 i.e. towards (away from) the YIG for negative (positive)
Fig. 2 for different temperatures are shown in Fig. 3 as
µ at detector 1 (detector 2), and consequently governs
m
redsymbolsforsampleseriesA.Withincreasingtemper-
the sign of the measured non-local ISHE voltage. Non-
ature d increases monotonically and seems to saturate
0 local SSE measurements as a function of the strip sepa-
around T =200K.
rationthereforeallowustomapoutthenon-equilibrium
Similar experiments as a function of temperature and magnon distribution in the YIG film. In particular the
strip separation were performed on devices from series characteristic length d for the sign change of µ can be
0 m
B and the critical strip separation extracted from these determined.
measurementsisincludedinFig.3asbluesquares. While In order to rationalize the measured temperature de-
the temperature dependence is much steeper, a qualita- pendence of d , the parameters governing the angu-
0
tivelysimilarincreaseofd0 withtemperatureisobserved lar momentum transfer across the YIG/Pt interface as
in both series. well as the magnon diffusion process need to be an-
Thischaracteristicsignchangeinthenon-localSSEin alyzed. It has been shown that the transparency of
YIG/Pt heterostructures above a particular separation the YIG/Pt interface, described by the effective spin-
d hasbeenpreviouslyobservedbyShanetal.11 atroom mixing conductance g , influences the magnon accumu-
0 s
temperature and was attributed to the spatial profile of lation and hence the sign-reversal distance d 11. For a
0
thenon-equilibriummagnonaccumulationµ intheYIG fully opaque interface (obtained using an Al O inter-
m 2 3
film, as shown schematically in Fig. 1 (b). Magnons are layer)whichsuppressesangularmomentumbackflowinto
thermally excited in the ferrimagnet due to Joule heat- the injector and therefore preserves a strong magnon de-
ing in the injector strip and diffuse vertically towards pletion, an increase of the sign-reversal distance d was
0
the GGG/YIG interface as well as laterally to the sam- observed11. Previous measurements of the MMR effect
ple edges. According to the model proposed by Shan et inaYIG/Ptheterostructureasafunctionoftemperature
4
haveshownthattheMMRsignaldecreaseswithdecreas- distance and temperature. The non-local SSE changes
ingtemperature5,consistentwithg ∝T3/2 aspredicted sign at a characteristic injector-detector separation d ,
s 0
by theory15,16. However, with this temperature depen- confirmingpreviousobservationsputforwardbyShanet
denceweexpectadecreasingtransparencyoftheYIG/Pt al.11. We furthermore observe a decrease of the charac-
interface with decreasing temperature, leading to an in- teristic separation d with decreasing temperature. Our
0
crease of d at low temperatures according to the model results suggest a complex dependence of the non-local
0
presented in Ref. 11. Since this is not consistent with SSE on interfacial transparency, magnon diffusion prop-
our experimental observations depicted in Fig. 3, the in- erties as well as phonon heat transport.
terface properties alone are not sufficient to describe the KG acknowledges N. Vlietstra for discussions and JC
temperature dependence of the non-local SSE. thanksS.Kauschkeforsamplepreparation. Thisworkis
Inadditiontotheinterfacialtransparency,themagnon financiallysupportedbytheDeutscheForschungsgemein-
diffusionlengthλ andthemagnonspinconductivityσ schaft through the Priority Program Spin Caloric Trans-
m m
determinethespatialdistributionofthenon-equilibrium port (GO 944/4, GR 1132/18, KL 1811/7) and the SFB
magnons in YIG. We extracted the magnon diffusion TRR 173 Spin+X, the Graduate School of Excellence
lengthfromtemperaturedependentMMRmeasurements Materials Science in Mainz (MAINZ) and EU projects
conducted in the sample series A and found an increase (IFOX,NMP3-LA-2012246102, INSPINFP7-ICT-2013-
of λ with decreasing temperature by about a factor of X 612759).
m
3 between 300K and 50K, following a 1/T dependence.
This is different from the temperature independent dif- 1A.V.Chumak,V.I.Vasyuchka,A.A.Serga, andB.Hillebrands,
fusion length reported by Cornelissen et al.17, who ex- “Magnonspintronics,”Nat.Phys.11,453–461(2015).
2A.V.Chumak,A.A.Serga, andB.Hillebrands,“Magnontran-
tracted λ (T) = const. together with a magnon spin
m sistor for all-magnon data processing,” Nat. Commun. 5, 4700–
conductivity σm vanishing at low temperatures. While (2014).
the detailed evolution of λ and σ with T thus must 3A.Khitun,M.Bao, andK.L.Wang,“Magnoniclogiccircuits,”
m m
be studied more systematically in future work, it is clear Jour.Phys.D:Appl.Phys.43,264005(2010).
4L.J.Cornelissen, J.Liu, R.A.Duine, J.B.Youssef, andB.J.
that these quantities depend on temperature. This im-
vanWees,“Long-distancetransportofmagnonspininformation
pliesthattheycanqualitativelyimpactthemagnondiffu-
in a magnetic insulator at room temperature,” Nat. Phys. 11,
sionprocessandconsequentlythenon-localSSE.Indeed, 1022–1026(2015).
thestrongdependenceofd ontherelativeamplitudesof 5S.T.B.Goennenwein,R.Schlitz,M.Pernpeintner,K.Ganzhorn,
0
g , λ and σ at a fixed temperature has been demon- M. Althammer, R. Gross, and H. Huebl, “Non-local magne-
s m m
toresistanceinYIG/Ptnanostructures,”Appl.Phys.Lett.107,
stratedbyShanetal. usingaone-dimensionalanalytical
172405(2015).
model for the spin Seebeck effect11. 6K. Ganzhorn, S. Klingler, T. Wimmer, S. Gepr¨ags, R. Gross,
Based on the available experimental data we can con- H. Huebl, and S. T. B. Goennenwein, “Magnon-based logic in
a multi-terminal yig/pt nanostructure,” Appl. Phys. Lett. 109,
clude that for a quantitative modeling of the non-local
022405(2016).
SSE the temperature dependence of both the angular
7L.J.CornelissenandB.J.vanWees,“Magneticfielddependence
momentumtransferacrosstheYIG/Ptinterfaceandthe of the magnon spin diffusion length in the magnetic insulator
magnondiffusioninYIGmustbetakenintoaccount. We yttriumirongarnet,”Phys.Rev.B93,020403(2016).
furthermore note that additional effects due to phononic 8K.Uchida,H.Adachi,T.Ota,H.Nakayama,S.Maekawa, and
E. Saitoh, “Observation of longitudinal spin-seebeck effect in
heat transport, known to be of importance for the local
magneticinsulators,”Appl.Phys.Lett.97,172505(2010).
SSE,cannotfullybeexcludedatthispointasasourceof 9A. Kehlberger, U. Ritzmann, D. Hinzke, E.-J. Guo, J. Cramer,
the non-local SSE-like signal. The local SSE originates G. Jakob, M. C. Onbasli, D. H. Kim, C. A. Ross, M. B.
from a finite difference of the effective temperature of Jungfleisch, B. Hillebrands, U. Nowak, and M. Kla¨ui, “Length
the magnon and phonon subsystems in YIG close to the scale of the spin seebeck effect,” Phys. Rev. Lett. 115, 096602
(2015).
YIG/Ptinterface,givingrisetoamagnonspincurrent18.
10E.-J.Guo,J.Cramer,A.Kehlberger,C.A.Ferguson,D.A.Ma-
While this difference may also influence the thermal sig-
cLaren, G. Jakob, and M. Kl¨aui, “Influence of thickness and
nalmeasuredatthenon-localdetectorstrip,itwasshown interface on the low-temperature enhancement of the spin see-
that the thermalization of the magnon and phonon sub- beckeffectinyigfilms,”Phys.Rev.X6,031012(2016).
systems takes place on a length scale λ of the order 11J. Shan, L. J. Cornelissen, N. Vlietstra, J. Ben Youssef,
mp T. Kuschel, R. A. Duine, and B. J. van Wees, “Influence of
of a few nm at room temperature19–21, due to a very ef-
yttriumirongarnetthicknessandheateropacityonthenonlocal
ficient (magnon conserving) magnon-phonon scattering. transport of electrically and thermally excited magnons,” Phys.
The effective temperature model described in Ref. 18 is Rev.B94,174437(2016).
thereforeapplicablemainlyinthelocallimit,i.e.closeto 12S. Pu¨tter, S. Gepr¨ags, R. Schlitz, M. Althammer, A. Erb,
R. Gross, and S. Goennenwein, Appl. Phys. Lett. accepted
the injector. The long-distance non-local limit however
(2016).
willbedominatedbydiffusingmagnonsandthemagnon-
13M. Schreier, N. Roschewsky, E. Dobler, S. Meyer, H. Huebl,
phonon scattering which does not conserve the number R.Gross, andS.T.B.Goennenwein,“Currentheatinginduced
ofmagnonsandcanreachalargerlengthscaleλ ofthe spinseebeckeffect,”Appl.Phys.Lett.103,242404(2013).
m
order of several µm16, as discussed above. 14M.Schreier,G.E.W.Bauer,V.I.Vasyuchka,J.Flipse,K.ichi
Uchida,J.Lotze,V.Lauer,A.V.Chumak,A.A.Serga,S.Dai-
In summary, we have measured the non-local SSE in a
mon, T. Kikkawa, E. Saitoh, B. J. van Wees, B. Hillebrands,
YIG/Ptheterostructureasafunctionofinjector-detector R.Gross, andS.T.B.Goennenwein,“Signofinversespinhall
5
voltagesgeneratedbyferromagneticresonanceandtemperature 18J.Xiao,G.E.W.Bauer,K.Uchida,E.Saitoh, andS.Maekawa,
gradientsinyttriumirongarnetplatinumbilayers,”Jour.Phys. “Theoryofmagnon-drivenspinseebeckeffect,”Phys.Rev.B81,
D:Appl.Phys.48,025001(2015). 214418(2010).
15S.S.-L.ZhangandS.Zhang,“Spinconvertanceatmagneticin- 19M. Schreier, A. Kamra, M. Weiler, J. Xiao, G. E. W. Bauer,
terfaces,”Phys.Rev.B86,214424(2012). R. Gross, and S. T. B. Goennenwein, “Magnon, phonon, and
16L.J.Cornelissen,K.J.H.Peters,G.E.W.Bauer,R.A.Duine, electrontemperatureprofilesandthespinseebeckeffectinmag-
and B. J. van Wees, “Magnon spin transport driven by the netic insulator/normal metal hybrid structures,” Phys. Rev. B
magnonchemicalpotentialinamagneticinsulator,”Phys.Rev. 88,094410(2013).
B94,014412(2016). 20S. R. Boona and J. P. Heremans, “Magnon thermal mean free
17L.J.CornelissenandB.J.vanWees,“Temperaturedependence pathinyttriumirongarnet,”Phys.Rev.B90,064421(2014).
ofthemagnonspindiffusionlengthandmagnonspinconductiv- 21J. Flipse, F. K. Dejene, D. Wagenaar, G. E. W. Bauer, J. B.
ityinthemagneticinsulatoryttriumirongarnet,”ArXive-prints Youssef, and B. J. van Wees, “Observation of the spin peltier
(2016),arXiv:1607.01506[cond-mat.mes-hall]. effect for magnetic insulators,” Phys. Rev. Lett. 113, 027601
(2014).