Surface acoustic wave-induced electroluminescence intensity oscillation in planar light-emitting devices

0.13 MB·English

by  Marco Cecchini

Surface acoustic wave-induced electroluminescence intensity oscillation in planar
light-emitting devices
∗
Marco Cecchini, Vincenzo Piazza, and Fabio Beltram
NEST-INFM and Scuola Normale Superiore, I-56126 Pisa, Italy
D. G. Gevaux, M. B. Ward, and A. J. Shields
5 Toshiba Research Europe Limited, Cambridge Research Laboratory,
0 260 Cambridge Science Park, Milton Road, Cambridge CB4 OWE, United Kingdom
0
2 H. E. Beere and D. A. Ritchie

n Cavendish Laboratory, University of Cambridge, Cambridge CB3 0HE, United Kingdom
a
J Electroluminescenceemissionfromsurfaceacousticwave-drivenlight-emittingdiodes(SAWLEDs)
7 is studied by means of time-resolved techniques. We show that the intensity of the SAW-induced
electroluminescenceismodulatedattheSAWfrequency(∼1GHz),demonstratingelectroninjection

] intothe p-typeregion synchronouswith the SAW wavefronts.
r
e
h
Surface-acoustic-wave(SAW) based devices are nowa- the acoustoelectric effects to planar systems where both
t
o days widely used for mobile and wireless applications, electron and holes are present and allow detailed studies
.
t as well as for satellite communications and military oftheeffectsoftheacousticmodulationinlight-emitting
a
applications[1]. The large-scale diffusion of sophisti- devices (LEDs).
m
cated communication systems stimulated a fast rise in In this Letter we report on time-resolved studies of
-
d the SAW-based device market, reaching a production theacoustoelectriceffectsinplanarLEDs. Wefabricated
n volume of several million devices per day[2]. More re- devices containing lateral n-p junctions and interdigital
o cently SAWs have attracted the interest of the semicon- transducersforSAWgeneration(SAWLEDs)andstudied
c ductorcommunityinview ofthe exploitationoftheirin- theopticalpropertiesofthedevicesinducedbytheacous-
[
teraction properties with two-dimensional-electron-gases tic perturbation. We monitored SAW-induced electro-

1 (2DEGs)embeddedinsemiconductorheterostructures[3, luminescence (EL) as a function of time and demon-
v 4]. SAWs propagating through mesas containing high- strated the presence of oscillations with the same period
6 quality 2DEGs indeed drive modifications on the 2DEG of the SAW (∼1 ns).
3 equilibrium state. Acoustic waves propagating along Devices were fabricated starting with a p-type
1
1 piezo-electric substrates are accompanied by potential modulation-doped Al0.3Ga0.7As/GaAs heterostructure
0 waves which can trap electrons in their minima and in- grown by molecular beam epitaxy, containing a two-
5 duce dc currents or voltages[5, 6, 7]. The discovery of dimensionalholegas(2DHG)withina20-nm-wideGaAs
0 the so-called acoustoelectric effect was followed by the quantum well embedded 70nm below the surface. The
/ proposal of innovative device concepts. Among these measuredhole density and mobility after illumination at
t
a Talyanskiiet al. proposedthe implementationof a novel 1.5K were 2.0×1011 cm−2 and 35000cm2/Vs, respec-
m current standard, demonstrating very precise acousto- tively.
- electriccurrentquantizationduetochargedragbySAWs The heterostructurewas processedinto mesas with an
d
througha quantum pointcontact[8, 9, 10, 11, 12]. Con-
annealed p-type Au/Zn/Au (5/50/150nm) Ohmic con-
n
trol over the constriction width allows very precise se-
o tact. The fabrication of the n-type region of the junc-
lection of the number of electrons packed in each SAW
c tion followed Ref.[15]. The Be-doped layer was removed
: minimum down to the single-electron-transport regime. from part of the mesa by means of wet etching (48s in
v
One of the most appealing applications proposed after
i H3PO4:H2O2:H2O = 3:1:50) and the sample was loaded
X the first report of the acoustoelectric quantized current
into a thermal evaporator for the deposition of a self-
r was to incorporate single-electron SAW pumps in pla- alignedNi/AuGe/Ni/Au(5/107/10/100nm)n-typecon-
a nar 2D electron/2D hole gas (n-p) junctions to fabricate
tact. After annealing, donors introduced by the n-type
high-repetition-rate single-photon sources.
contactprovideconductionelectronswithinthewellthus
Very recently two different groups demonstrated one creatinganelectrongasbelowthemetalpad,adjacentto
of the main building blocks of such source, that is the the 2DHG (inset of Fig1(a)). The n-contact was shaped
possibilitytoswitchontheelectroluminescenceofplanar as a thin stripe placed perpendicular to the SAW propa-
n-p junctions by applying SAWs[13, 14]. From a more gation direction, 250µm away from the p-contact.
fundamental physics point of view these results extend SAWs propagating along the (0¯1¯1) crystal direction
were generated by means of an interdigital transducer
(IDT) composed of 100 pairs of 200µm-long Al fingers
with 3µm periodicity (∼1GHz resonance frequency on
∗Electronicaddress: [email protected] GaAs). Transducers were fabricated at a distance of
2
EL measurements at low temperature (5K). Spectra as
(a)
10-3 5 K a function of bias were collected by a cooled CCD af-
ter spectral filtering by a single-grating monochromator.
TheELspectrareportedinFig.1(b)showamainpeakat
10-5
818.7nm (full width half maximum (FWHM) of 1.8nm)
RT
originatingfromradiativerecombinationwithin the QW
)
(A 10-7 p-type Be-doped layer andasecondarypeakat831.2nm(FWHM3.6nm)which
I contact n-type originates from carbon impurities included in the het-
contact erostructure material during growth[16]. The origin
10-9 (cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0) (cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1) of such peaks was also verified by photo-luminescence
Holes Electrons
measurements[17]. Light intensity, calculated by inte-
GaAs QW
gration over the main EL peak from 810nm to 826nm,
as a function of the forwardbias applied to the junction
0 1 2 3 4
Forward Bias (V) reflects the rectifying behavior of the IV curves (inset of
Fig.1(b)), showing a detection threshold of ∼1.65V.
(b) 1.8 V TheSAWresonancefrequencywasdeterminedbymea-
u.) 1.7 V suringthepowerreflectedbythetransducerasafunction
b. u.) 1.6 V of excitation frequency. The transducer frequency re-
ar b. sponse displayedat 5K a pronounceddip at 987.5MHz,
( ar
y y ( whichisconsistentwiththeperiodicityofthetransducer,
sit nsit and had a FWHM of 2.4MHz.
Inten EL Inte T = 5 K driTvheeeltercatvreolnlisnginetloectthreic2fiDelHdGaswsohceinatethdewjiutnhcStiAoWn siscbain-

L 1.6 1.7 ased just below threshold [13, 14]. Indeed, we observed
E
Forward Bias (V) the light versus voltage curve to shift to lower biases by
∼10meVinpresenceofSAWs(Fig.2(a)). Themaximum
shiftwasobservedat-10dBmpowerlevel. Athigherlev-
800 810 820 830
els the increased electron extraction is counterbalanced
Wavelength (nm)
by the spatial separation of electrons and holes trapped
respectively in SAW minima and maxima, leading to a
FIG. 1: (a) Current-voltage characteristics from room tem-
progressivesuppression of the EL signal.[14]
perature down to 5K. Inset: fabrication scheme of the n-p
The emission spectra were changed by the SAW only
lateral junction. (b) Electro-luminescence spectra at differ-
in intensity and no spectral shift of the main peaks was
ent forward biases around the emission detecting threshold.
observed (Fig.2(b)). Light-intensity data as a function
Inset: Light-voltage characteristic of the planar LED at T =
5K. of IDT excitation frequency demonstrate that emission
intensification occurs only within the transducer pass-
band, as shown in the inset of Fig.2(b).
800µmfromthemesabyelectron-beamlithography. The We also studied the SAW-induced EL signal by time-
width of the n-type contact stripe (2µm) was chosen of resolved measurements. To this end the EL signal was
the same order of magnitude as the SAW wavelength spectrally filtered by a triple grating monochromator
(3µm)inordertolimitSAWdampingduetothemassive and detected by a single-photon APD module (Perkin
metalization and SAW diffusion originating from non- ElmerSPCM-AQR-16). Inordertoobtaintime-resolved
uniform penetration of the Ohmic contact and inhomo- ELtracesweusedtime-correlatedphoton-countingtech-
geneities in the etched region. niques. The signal from the SPCM was directed toward
SeveralSAWLEDswerefabricatedandstudied,andall theSTARTinputofaBecker&HicklSPC-600computer
of them showed qualitatively similar characteristics. All board,whiletheSTOPinputwasdrivenbyasignalsyn-
the data shown in the following refer to one representa- chronous to that used to generate the SAW. A typical
tivedevice,forwhichthetransportandopticalproperties data trace is shown in Fig.3(a). EL oscillations with
ofthejunction,aswellastheefficiencyofthetransducer, period of about 1ns are present in the curve, reflecting
will be systematically analyzed. the presenceofthe SAWmodulatingthe junctionpoten-
The SAWLED was first electrically tested at low tem- tial. TheseoscillationsaremoreclearlyvisibleinFig.3(c)
perature to verify the proper formation of the lateral where a magnified view of Fig.3(a) is displayed with a
diode. Current-voltage curves (IVs) at different temper- smootheddatatrace. Thiseffectisfurtherhighlightedin
atures(fromroomtemperaturedownto5K)showedrec- the frequency domain of the signal. Fig.3(b) reports the
tifyingbehaviorandintersectedat∼1.5V(seeFig.1(a)). Fourier transform of the EL signal. A single peak at the
This value corresponds to the diode conduction thresh- SAWfrequency clearlydominatesthe Fouriertransform.
oldanditis consistentwiththe valueexpectedforGaAs Time-correlated measurements were made at different
n-pjunctions. Emissionpropertieswerecharacterizedby biasesappliedtothejunction. Theoscillationamplitudes
3
) Time (ns)
.
u (a)
. NO SAW 10 15 20 25 30 35 40
b
r SAW (-10 dBm) (a)
a s
sity ( Count800 ( b)
n %)1.8
te 6 > (1.6
n C
L I T = 5 K e C/<1.4
E 1.62 1.64 1.66 1.68 1.70 1.72 ud 4 D1.2
Forward bias (V) mplit 1.7 1.8
Forward bias (V)
A
2
) (b) Bias = 1.675 V
u. T = 5 K

b. u.)
sity (ar ensity (arb. NSAOW SAW 0.5 F1.r0equen1.c 5y (GH2z.0) 2.5 (c)
n nt 880
nte EL I 980 990 1000 860
I RF frequency (MHz) s
L nt
E u 840
o
C
800 810 820 830 820
Wavelength (nm)
800
FIG. 2: (a) Light-voltage characteristic of the planar LED 42.0 42.5 43.0 43.5 44.0 44.5
without SAW (dashed line) and with SAW (solid line). Time (ns)
SAW power and frequency were -10dBm and 987.5MHz re-
spectively. (b) Electro-luminescence spectra without SAW
FIG.3: (a)ELtimeevolutioninpresenceofSAW(987.5MHz,
(dashedline)andinpresenceofaSAW(-10dBm,987.5MHz)
-10dBm), at T = 5K and for a forward bias of 1.85V. (b)
(solid line). Inset: electro-luminescence as a function of RF-
Fouriertransform ofthetoppanelEL signal. Inset: Normal-
frequency (-9 dBm) at T = 5K. The junction was forward
izedoscillationintensityasafunctionoftheforwardbias. (c)
biased with 1.8 V.
ZoomofthetimeevolutionofSAW-inducedELshowedin(a)
superimposed onto an FFT smoothing.
werecalculatedfromtheFouriertransformsandnormal-
ized by the corresponding mean EL intensities. In the recombining at different times depending on their posi-
inset ofFig.3(b)we reportthe normalizedoscillationin- tion along the junction. The resulting oscillations are
tensity as a function of the forwardbias. The oscillation therefore averaged out, while the EL mean-intensity in-
amplitude is around 1.5% of the signal and remains al- creasestillremainspresent. Webelievethatfurtheropti-
most unchanged at different bias voltages, as expected mizationofthedevicegeometrywillleadtohighcontrast
for an exponential increase of the EL close to the con- oscillations and is currently being investigated.
duction threshold. Since the recombination time in our In conclusion, we fabricated n-p lateral junctions with
system is much less than 1 ns, determined by means of interdigital transducers and studied the effect of SAWs
photoexcited-carrier lifetime measurements, the reason on the device time-resolved emission properties. SAW-
for the observed small oscillation amplitude originates inducedemissionwasfirstcharacterizedbyspectralmea-
from other limiting factors. Ideally the SAW wavefronts surements andlight-voltagemeasurements. Electrolumi-
would get to the thin n-type contactwithout any distor- nescencewasobservedtoincreaseinintensityinthepres-
tion and perfectly parallel to the contact itself. But real ence of SAWs when the diode was biased near the con-
devices introduce deviations from the picture described duction threshold. Time-resolved measurements of the
above. The presence of the mesa edges and of defects electroluminescence in the presence of SAWs were car-
alongthe SAW pathintroducedduring the mesa etching ried out at severalbias voltages,demonstrating modula-
process constitute scattering centers which lead to mod- tion of the light intensity at the frequency of the SAW
ification of the acoustic wave-fronts. In addiction, an- (∼1GHz). Theamplitudeoftheoscillationwasfoundto
nealed Ohmic contacts have a strong corrugation. Dur- be∼1.5%ofthetotallightintensity,almostindependent
ingtheannealingprocessmetaldiffusionmakesthejunc- from the bias applied to the junction.
tion edgesirregularonthe scale ofthe SAW wavelength. This work was supported in part by the European
These two factors lead to electrons being extracted and Commission through the FET Project SAWPHOTON
4
and by MIUR within FISR “Nanodispositivi ottici a pochi fotoni”.
[1] C.K.Campbell,Surfaceacousticwavedevicesformobile [10] J. Ebbecke, G. Bastian, M. Bl¨ocker, K. Pierz, and F. J.
and wireless communications (Academic Press, 1998). Ahlers, Appl. Phys.Lett. 77, 2601 (2000).
[2] C. C. W. Ruppel and T. A. Fjeldly, Advances in sur- [11] V. I. Talyanskii, J. M. Shilton, M. Pepper, C. G. Smith,
face acoustic wave technology, systems and applications C.J.B.Ford,E.H.Linfield,D.A.Ritchie,andG.A.C.
(World ScientificPublishing, 2000). Jones, Phys. Rev.B 56, 15180 (1997).
[3] A. Wixforth, J. P. Kotthaus, and G. Weimann, Phys. [12] J.M.Shilton,V.I.Talyanskii,M.Pepper,D.A.Ritchie,
Rev.Lett. 56, 2104 (1986). J. E. F. Frost, C. J. B. Ford, C. G. Smith,and G. A.C.
[4] R.L.Willett,R.R.Ruel,K.W.West,andL.N.Pfeiffer, Jones, J. Phys.: Condens. Matter 8, L531 (1996).
Phys.Rev.Lett. 71, 3846 (1993). [13] T. Hosey, V. Talyanskii, S. Vijendran, G. A. C. Jones,
[5] A.Esslinger,R.W.Winkler,C.Rocke,A.Wixforth,J.P. M. B. Ward, D. C. Unitt, C. E. Norman, and A. J.
Kotthaus,H.Nickel,W.Schlapp,andR.Lsch,Surf.Sci. Shields, Appl. Phys.Lett. 85, 491 (2004).
305, 83 (1994). [14] M.Cecchini, G.D.Simoni,V.Piazza, F.Beltram, H.E.
[6] A. Esslinger, A. Wixforth, R. W. Winkler, J. P. Kot- Beere, and D. A. Ritchie, Appl. Phys. Lett. 85, 3020
thaus, H. Nickel, W. Schlapp, and R. Lsch, Solid State (2004).
Commun. 84, 939 (1992). [15] M.Cecchini,V.Piazza,F.Beltram,M.Lazzarino,M.B.
[7] J. W. M. Campbell, F. Guillon, M. D’Iorio, Ward,A.J.Shields,H.E.Beere,andD.A.Ritchie,Appl.
M. Buchanan, and R. J. Stoner, Solid State Com- Phys. Lett. 82, 636 (2003).
mun.84, 735 (1992). [16] B. Hamilton, in Properties of Gallium Arsenide, edited
[8] J. Cunningham, V.I.Talyanskii, J. M. Shilton, M. Pep- by M. R. Brozel and G. E. Stillman (INSPEC, London,
per,A.Kristensen,andP.E.Lindelof, Phys.Rev.B 62, England, 1996).
1564 (2000). [17] PLspectrawereobtainedbyexcitationofaregionofthe
[9] J. Cunningham, V.I.Talyanskii, J. M. Shilton, M. Pep- mesa with a red-light laser source (653nm).
per, M. Y. Simmons, and D. A. Ritchie, Phys. Rev. B
60, 4850 (1999).