Piezoreflectance characterization of double‐barrier resonant tunneling structures

Abstract

The piezoreflectance technique has been used to optically characterize resonant tunneling structures that utilize isolated single quantum wells. The heavy‐ and light‐hole transitions associated with the quantum wells were prominent in the spectra of samples with barrier widths ranging from 50 to 34 Å. Their spectral positions depended not only on quantum well and barrier thicknesses, but also significantly on the amount of carrier confinement produced by barrier height. Furthermore, variations in the magnitude of impurity transitions could be observed in the spectra of different samples.

Full text

Page 1

Piezoreflectance characterization of doubleftbarrier resonant tunneUng
structures
R. L. Tober
u.s. Army Laboratory Command, Harry Diamond Laboratories, 2800 Powder llJill Road, Adelphi,
A1aryland 20783-1197
J. Pamulapati, J. E. Oh, and P. K. Bhattacharya
Solid State Electronics Laboratory and Center jor High Frequency Microelectronics, Department of
Electrical Engineering and Computer Science. University oflWichigan. Ann Arbor, Michigan 48109-2122
(Received 4 February 1988; accepted for publication 24 June 1988)
The piezorefiectance technique has been used to optically characterize resonant tunneling
structures that utilize isolated single quantum wells. The heavy- and light-hole transitions
associated with the quantum wells were prominent in the spectra of samples with barrier
widths ranging from 50 to 34 A. Their spect.ral positions depended not only on quantum weB
and barrier thicknesses, but also significantly on the amount of carrier confinement produced
by barrier height. Furthermore, variations in the magnitude of impurity transitions could be
observed in the spectra of different samples.
In spite of the burgeoning interest in double-barrier res-
onant tunneling devices (see, for example, Refs. 1-5), no
suitable techniques have emerged that can probe their isolat-
ed quantum well structures. The highly absorbing doped
GaAs layers that cover the double-barrier regions permit
electrical contact, but prohibit photoluminescence, electro-
reflectance, or photoreflectance measurements of the energy
levels that aid in the tunnelbg process. This has necessitated
the use of irregular characterization techniques. The growth
parameters of these resonant tunneling devices have hereto-
fore only been indirectly inferred from electrical measure-
ments (i.e., current-voltage)5 or extrapolated from optical
measurements performed on multiple quantum wells or
near-surface structures.6
This letter reports on the viability of the piezorefiec-
tance (PzR) technique for characterizing resonant tunnel-
ing devices that utilize isolated quantum well structures. The
optical spectra 110t only indicate resonant energy levels, but
clearly demonstrate their dependence on growth param-
eters. Moreover, the nondestructive nature and sensitivity to
impurity levels establishes PzR as a valuable technique for
characterizing double-barrier heterostructures prior and
subsequent to device fabrication. It should be emphasized
that although the piezoreflectance technique has been used
in the past to characterize undoped single or muItiquantum
wen structures,7 it has never been used to characterize struc-
tures such as resonant tunneling diodes. Therefore, the in-
formation presented in this letter should benefit those in-
volved in the design, fabrication, and characterization of
these devices.
AU four of the molecular beam epitaxially grown dou-
ble-barrier resonant tunneling samples used in these experi-
ments had 0.5 {tm of Si doped n j. -GaAs grown over the
double-barrier regions. Two of them had 45 L'\ GaAs wells:
one (sample UMl) with 34 A AlAs barriers, the other (sam-
ple UM2) with 45 A AlAs barriers. The second set of sam-
ples (grown at a separate facility) had 50 A Alo.25 GaO.75 As
barriers: one with a 40 A wen (PEl), the other a 50 A wen
(PE2).
The piezoreflectance apparatus closely resembled that
used elsewhere,7,B and therefore, only relevant details will be
presented here. Sample substrates were glued to lead-zircon-
ate-titanate piezoelectric transducers (the cyanoacrylate es-
ter glue could easily be removed with acetone without dam-
aging the sample). A 400 Hz 185 v'n" square wave applied
to the transducer produced strains (<: 10-5) :mfficient to
modulate the electronic structure of the double-barrier sam-
ples, yet small enough to ensure a linear relation to flR IR.
Radiation from a tungsten-halogen lamp was dispersed
through a 114m monochromator and focused toa ~ 2 5 0 1 l m
spot on the sample. The modulated reflectance signal was
measured using a silicon p-i-n photodiode and lock-in ampli-
fier, then normalized digitally to the dc reflectance.
Figure 1 shows the differences and similarities between
typical PzR spectra obtained from the Physical Electronics
(PE) (dashed) and U ni versity of Michigan (UM) (solid)
samples. The PzR spectra of the PE samples always had
features (labeled 1) attributed to impurity acceptor levels.
The UM samples had similar impurity transitions (1'), but
with significantly smaller magnitUdes, Transitions involving
0.0
(D.O ')-
- - - - - - ~ t ~ - - - - - - - - ~ ~ - - f - - - - ' t
t'
Z'
-1.OL-----;1"'.a-----:;1"".'-----;1;';;.S,------:1"".6:------,1'"'.7;-------'
3' 4'
.m
E N E R ~ Y , . ~ )
FIG. 1. Typical worn-temperature piczoreflectance spectra of the FE
(dashed) and UM (solid) samples.
883
AppL Phys. Lett. 53 (10),5 September 1988 0003-6951/88/360883-03$01.00
@ 1988 American Institute of PhySics
883

Page 2

carbon acceptor levels have previously been observed with
low-temperature PzR measurements on MBE quantum weB
samples at energies ~
ever, the impurity transitions shown in Fig, 1 occurred ~
me V below the GaAs feature (2 and 2'), implying that they
resulted from Si acceptor levels,!) The difference in the mag-
nitudes of these features suggested that the PE sample:; were
grown under conditions that enhanced the amphoteric na-
ture of Si, 10 All PzR spectra obtained from these resonant
tunneling samples had two features (labeled 3 and 4 or 3' and
4' in Fig, 1) attributed to the E Ill! heavy-hole and E l1L
light-hole transitions associated with the isolated single
quantum wells,
Comparisons of PzR spectra obtained from a multiple
quantum well sample (50 A GaAs wells and 150 A
Alo.3 GaO.7 As barriers) with that from the PE2 sample (50 A
GaAs weB and 50 A Alo25 GaO.75 As barriers) aided in identi-
fying resonant energy levels associated with the double bar-
riers, The maxima in the former spectra were previously
shown to occur at the energies of transitions to excitonic
states, II However, since the relative position of maximum
with respect to the actual transition energy and the spectra
line shape depend on the cxcitonic character of the final
state, 12, 13 extrema in the spectra of resonant tunneling sam-
ples were only used as a first approximation to the energies of
the E ! tH,!. transitions,
The PzR spectra of Fig, 2 demonstrate the energy de-
pendence of the E IIH and E [IL transitions on quantum well
width, The n = 1 Hght- and heavy-hole transitions associat-
ed with the 40 A well not only occurred <It energies greater
than those from the 50 A well, but also their energy separa-
tion increased.
A reduction in barrier width (while maintaining barrier
height and well width) reduced carrier confinement, thereby
decreasing the energy of the }1Cavy- and light-hole transi-
tions as indicated in Fig, 3, Both of these PzR spectra were
obtained from samples with 45 A wells, but only had 34 A
barriers (solid curve); the other had 45 A barriers (dashed
curve), Also, note the increased Iinewidth of the E IlL fea-
20 me V below the GaAs level. 7 How-
35
-1.0
-1.5 L_, __ ..L. ___ -.L._._..L. ___ . _ .. 1-. __ _
1M
,~
1."
El1H
1,506
Ej1l
1.527 :
1~
I.
EUERGY (.V)
FIG, 2, PzR spectra of the FE samples having 40 A (dashed) and 50 A
(solid) wells, Notethatthe increase in confinement shifted the n = 1 transi-
tions toward greater energies,
884
Appl. Phys. Lett, Vol. 53, No, 10, 5 September 1988
1.65
< N E ~ G V (eV)
1,73
1.75
FIG, 3. PzR spectra clearly show the effects of barrier width on carrier
confinement. The samples had 34 A (solid) and 45 A (dashed) AlAs bar-
riers,
ture and the shoulder on the low-energy side of the E I]J{
feature in the spectrum of the 34 A barrier sample. Addi-
tional features such as these have previously been ascribed io
interface irregularities that occur during growth,"
Figure 4 illustrates the shift in energy of the n = 1 levels
that accompanied an increase in barrier height The E [ !/I,L
features in the UM2 spectrum (AlAs barriers) occurred at
energies ~
(Alo25 Gao 75 As barriers), The energy split between the
light- and heavy-hole transitions increased as welt The mi-
nor difference in the well and barrier width bet ween the PE 1
and UM2 samples only decreased the separation of the cor-
responding n = I transitions compared to that which would
have occurred had only their barrier height differed, These
spectra, therefore, demonstrate the significance of barrier
height to carrier confinement.
In conclusion, the PzR technique has proven useful for
directly determining energy levels associated with quantum
wells deeply buried ( ~ O , 5 plTI) below regions of heavily
100 meV greater than those of the PEl sample
t ~ · u ~
1.526
j
FIG, 4, PzR spectra illustrating the effects of barrier height on carrier con-
finement. The samples had AIG.25 Ga.175 As (dashed) and AlAs (solid) bar-
riers.
Tober etal,
884

Page 3

doped GaAs. The sensitivity of PzR to subtle differences in
growth parameters, impurity levels, and surface inhomo-
geneities mah'S it useful for predicting device performance
prior to fabrication. Furthermore, the dependence of PzR
Hne shapes on excitonic contrihutions provides a means to
study carrier confinement and system dimensionality.
The portion of this work completed at the University of
Michigan was supported hy 1he U.S. Army Research Office
(URI Program) under contract No. DAA03-87-K-0007.
Ie. Sollner. W. D. Goodhue, P. E. Tannenwakl, C. D. Parker, ana D. D.
Peck, Appl. Phys. Lett. 43, 588 (1983).
'A. R. BOllndoi, R. T. Collins, T. C. Mcgill, R. D. Burnham, and F. A.
Ponce, App\. Phys. Lett. 6, 2R5 (lgB5).
'J. She'wchuck, P. C ChaptlJ, P. D. Coleman, W. Kopp, R. Fischer. and H.
Morko<;. AprL Phys. Lett. 46, 50g (19R5).
885
Appl. Pi1YS. Lett., Vol. 53, No.1 0,5 September 1988
'E. E. Mendez, W. 1. Wang, B. Ricco, and L. Esaki, Aprl. Phys. Lett. 47,
415 (1985),
'M. A. Reed, J. W. Lee, R K. Aldert, and A. E. Wetzel, 1. Mater. Res. 1,
337 (1986).
'M.Tsuchiya, T. Matsusue, and H. Sakaki, Phys. Rev. Lett. 59, 2356
( 1987).
7y. R. Lee, A. K. Ramclas, F. A. Chambers, J. M. Meese, and L. R. Ram
Mohan, AppL Phys. 50, 600 (191$7).
'M. Garfinkel, 1. J, Tiemann, and W. E. Engler, Phys. Rev. 148, 695
(1966).
"A. M. White, P. J. Dean, D. J. Ashen, 1. B. Mullin, M. W<'bh, B. Day, and
P. D. Green. J. Phys. C 6, L243 (1973).
10K. Ploog, A. Fisher, ana H. Kunze!, J. E!ectrochem. Soc. 128, 400
(1981).
I 'R, L Tober, Thomas F. Bogg<:ss and Arthur L. Smirl (unpublished).
"n. Pistoulet, H. M,HhiclI, J. Camasse1. and D. Auvergne, Phys, Status
Solidi 39, 363 (1970).
"J. Camassel, D. Auvergne, H. Mathieu, and 13. Pistoulet, Phys. Staws
Solidi 42, 147 (InO).
Tober eta!.
885