- Email: [email protected]
Modulation-doped HgCdTe quantum well structures and superlattices grown by photoassisted molecular beam epitaxy
Journal of Crystal Growth 101 (1990) 23—32 North-Holland
23
MODULATION-DOPED HgCdTe QUANTUM WELL STRUCTURES AND SUPERLATHCES GROWN BY PHOTOASSISTED MOLECULAR BEAM EPITAXY J.F. SCHETZINA, J.W. HAN, Y. LANSARJ, N.C. GILES, Z. YANG, S. HWANG and J.W. COOK, Jr. Department of Physics, North Carolina State University, Raleigh, North-Carolina 27695-8202, USA
and N. OTSUKA School of Materials Engineering Purdue University, West Lafayette, Indiana 47907, USA
We report the results of the first systematic study of modulation doping in HgCdTe. The doped multilayers were grown by means of photoassisted molecular beam epitaxy, in which the substrate is illuminated during the film growth process to enhance dopant activation. Indium was used as the n-type dopant and arsenic was used as the p-type dopant. More than forty modulation-doped samples were prepared and studied.
1. Introduction Precise control of the electrical properties of HgCdTe through the addition of substitutional impurities is an essential requirement for the
At North Carolina State University (NCSU), we have successfully circumvented this fundamental problem by employing modulation doping techniques to produce stable p-type alloys of Hg1 _5Cd~Te(x = 0.18—0.26) [3] by means of pho-
fabrication of infrared detectors based on in-situ grown p—n junctions or p-on-n heterojunctions, the most important area of application for HgCdTe at present. However, growth of stable p-type HgCdTe by MBE (a mercury-deficient growth process) has not been accomplished to date. Boukerche et al. [1] have recently reviewed the problems associated with the use of both Group I elements (Li, Ag, Na) and Group V elements (Sb, As) as potential p-type dopants in HgCdTe. It appears at present that none of these elements are suitable for use as p-type dopants in MBE-grown HgCdTe. The Group I elements lack stability they are fast diffusers in HgCdTe. Thus, one must question the long term viability of HgCdTe homojunctions of heterojunctions which contain these dopants [2]. The Group V elements appear to be amphoteric and often act to produce n-type material due, perhaps, to the Hg-deficient growth conditions which prevail during epitaxial growth of HgCdTe by MBE [1].
toassisted MBE [4]. n-Type modulation-doped samples of Hg1_~Cd~Te (x = 0.18—0.26) have also been successfully prepared [5]. In our initial studies, 8-doped Hg015Cd085Te barrier layers of thickness Lb 50 A were alternated with Hg1....5Cd5Te (x = 0.18—0.26) matrix layers of thickness L~ 1000 A. This produced a new material a quantum alloy of HgCdTe the properties of which are controllable on an atomic basis. Recently, we have completed the growth of new modulation-doped quantum well structures and superlattices of HgCdTe by means of photoassisted MBE technique. In this paper, details of the MBE growth experiments are given along with a discussion of the structural, electrical and optical properties that these interesting new HgCdTe quantum structures exhibit. Modulation doping involves the transfer of carriers (electrons or holes) from a substitutionally doped layer (modifier layer) to an adjacent material having a smaller band gap (matrix layer).
—
0022-0248/90/$03.50 © 1990
—
—
—
—
Elsevier Science Publishers By. (North-Holland)
—
24
J.F. Schetzina et al.
/ Modulation-doped HgCdTe QWstructures and SLa grown by PAMBE
In 1969, in their pioneering work on superlattices, Esaki and Tsu [6] proposed a selectively~doped heterojunction structure for enhanced carrier transport parallel to the interface. Independently in 1978, Dingle et al. [7], using n-type AIGaAs— GaAs heterostructures, first demonstrated that such enhanced carrier transport does, indeed, occur and first evoked the name “modulation doping”. Since then, modulation-doped structures composed of Ill—V semiconductor layers have been studied in detail, and the properties that these novel structures possess have subsequently been exploited in many planar device applications [8]. In 1987, n-type modulation-doped CdMnTe: In—CdTe quantum well structures were successfully prepared for the first time by photoassisted MBE at NCSU [9].
2. Experimental details The modulation-doped HgCdTe samples were grown in a Hg-compatible MBE system designed and built at NCSU. A description of this MBE system together with the techniques that have been developed to grow HgCdTe by conventional MBE are given in an earlier publication [10]. The Hg-MBE system was modified for photoassisted MBE growth experiments by replacing one of the MBE ovens on the main source flange with a UHV Pyrex window. The broad-band output (458—514 nm) from a Spectra-Physics model 201605 argon ion laser was beam-expanded and passed through the window to allow uniform illumination of the substrate during film growth. The laser power densities employedsurface ranged fromsubstrate 40—100 2 at the substrate so that mW/cm effects due to the incident laser beam are heating negligible. In our most recent work, p-type modulationdoped HgCdTe superlattices were grown by alternately depositing Hg 0 3Cd07Te barrier layers that were heavily doped with arsenic with undoped Hg078Cd022Te layers. In this way, superlattices consisting of 200 double layers were prepared. The thickness of the Hg0 3Cd07Te: As doping layers was Lb 50 A in each of the superlattices, with the As dopant uniformly distributed (no setback). —
For the small-band-gap well layers, Hg0 78Cd022Te of thickness L2 = 50—104 A was used in the various superlattice growth experiments. To prepare n-type modulation-doped HgCdTe superlattices, the wide-band-gap Hg03Cd07Te barrier layers were doped with indium rather than arsenic. pType and n-type modulation-doped superlattices consisting of 200 double layers of Hg0 3Cd07Te— HgTe have also been prepared and studied. Preliminary growth experiments of n-type modulation-doped heterostructures of HgCdTe which contain an undoped spacer layer have also been completed. These multilayered structures consist of a 200 A doping layer of Hg03Cd07Te: In onto which is deposited an undoped Hg0 3Cd07Te spacer layer followed by a 2000 A layer of Hg078Cd022Te. The ,spacer layer thickness was varied from 50—420 A in the various MBE growth runs. All of the modulation-doped HgCdTe samples were grown using lattice-matched (100) Cd096 Zn0~Tesubstrates. The substrates were polished and etched using standard techniques [11]. Immediately prior to film growth, the substrates were preheated at 300°Cfor 10 mm in the MBE chamber to drive off residual impurities and to insure a stoichiometric growth surface. A substrate temperature of 170°Cwas used throughout the HgCdTe film growth experiments. The structural perfection of the MBE-grown multilayers was assessed by means of doublecrystal X-ray rocking curve experiments performed using a Blake double-crystal diffractometer equipped with a Philips X-ray generator (Cu Ka X-rays). This instrument has a rocking curve resolution of 1 cross-section arc sec. Vertical transmission electron microscopy (TEM) micrographs were obtained at Purdue University using a JEOL model JEM 2000-EX microscope. Cross-sectional specimens for TEM studies were prepared by mechanical —
grinding followed by ion milling. During the ion milling operation, the specimens were cooled with liquid nitrogen to avoid intermixing of layers. Electrical characterization experiments consisted of Van der Pauw—Hall effect measurements performed on each modulation-doped sample over the temperature range 20—300 K.
J.F. Schetzina et aL
/ Modulation-doped HgCdTe QWstructures and
Infrared photoluminescence (PL) measurements at 4.2 K were completed on selected sampies using a Nicolet 60 SXR Fourier-transform infrared spectrometer (FTIR). In these experiments, the PL signal was excited using the 1.06 ~smemission from a Nd YAG laser focussed to a spot on the sample surface to2.give a power density of approximately 200 W/cm 3. Results and discussion 3.1. Modulation-doped HgCdTe with periodic 8doped barrier layers
p., .
The periodic nature of the modulation-doped HgCdTe samples containing 6-doped Hg 015
SL~grown by PAMBE
25
Cd 085Te barrier layers is clearly manifested by the vertical cross-section TEM micrographs obtained. Fig. la shows a cross-sectional bright field image of sample B15B. As seen from the figure, this sample consists of a matrix of Hg0 74Cd026Te in which twenty 6-doped Hg015Cd085Te layers of thickness Lb A are located periodically and separated by = 51.8 1130 A of the matrix material. The entire modulation-doped structure is shown in cross-section. Note the high structural quality of the epitaxial layers and their high degree of regularity. This is illustrated in more detail in the high resolution dark field image of B15B shown in fig. lb. The structural perfection of the modulationdoped samples is also illustrated by the X-ray —
diffraction spectra which they exhibit. This is
Q2J~ Fig. 1. (a) Bright field vertical cross-section TEM image of p-type modulation-doped structure B15B. The CdZnTe substrate is to the lower right. Twenty 51.8 A thick layers of Hg0 15Cd055Te doped with As appear as a series of parallel bright lines in the multilayer. Note that the superlattice is essentially dislocation-free. (b) High resolution dark field image of B15B showing seven 51.8 A thick layers of Hg0 15Cd035Te doped with As (bright parallel lines). Note the interfacial sharpness of this p-type modulation-doped structure.
26
J.F. Schetzina et aL o
/ Modulation-doped HgCdTe QW structures and SLs grown by PAMBE 0
..—.
Hg 1_~Cd~Te—Hg 1~Cd~TeSuperlattice — -~
Sample B15B T = + 170 L5 = °C 1180
~
Hg1_~Cd~Te—Hg1_~Cd~Te Superlattice
A
‘~
SBRC CdZnTe Substrate
A
________________________________ .
~ 0
Rocking
Sample B2OA T5=+ L5 170°C = 860
r
A
SBRCCdZnTe
Double crystal X ray
p 2
~
T
Ze, —~
o ‘-‘
_7~6 — —1000
__
+
+4
I
0 ANGLE (arc sec)
9
I
I
•
+5
I
0
1000
-,
51
UA
Sample B15B 170°C
+3
___________________ I
—1000
0 ANGLE (arc see)
2
I
I
+4 I
1000
I
•
Hg1 _~Cd~Te_Hg1_~Cd~Te Superlattice
Hg1~Cd~Te—Hg1_~Cd~Te Superlattice
~ a
2
Sample + L5 B2OA = 860 T5= 170°C
.
~
A
SL Principal Peak
>,
A
Double L5 c~t~ X—ray = 1180
~ ~
/U\ ~SL 30Principal arc sec Peak =
z ~
Rocking Curve
mm ~
Beam Cross Section: ~
_____
o —400
1mm x 1
—200
~M~400~
I-
Substrate \\~3BR~CdZTe FWHM(400) = 20 arc sec
Beam Cross Double Rocking crystal Curve Section: x — ray
5-
x im 0
2 arc sec ~- SBRC CdZnTe Substrate FWHM (400)
‘I
0 200 400 —400 —200 0 ANGLE (arc sec) ANGLE (arc sec) Fig. 2. Double-crystal X-ray diffraction rocking curves for samples B12B and B2OA.
shown in fig. 2 for samples B15B and B2OA. At the left is shown the X-ray diffraction rocking curve of B15B (consisting of 51.8 A modifier layers separated by 1130 A thick Hg0 74Cd026Te matrix layers). Since the modifier layers are placed periodically, the sample is thus a superlattice. This is reflected by the occurrence of X-ray satellite peaks. Seven orders of satellites are shown in spectrum of B15B, attesting to the excellent structural quality that has been achieved. Note that the spectrum was obtained using a double-crystal diffractometer with the detector full open (rocking curve configuration) and that the FWHM of the —
16 arc sec 200 400 =
main superlattice peak and each of the sateffites is 30 arc sec. At the right in fig. 2, corresponding double-crystal rocking curves for an n-type modulation-doped sample (B2OA) are shown. Again, multiple orders of satellites are observed with the SL principal peak having a FWHM = 22 arc sec. These spectra are comparable in quality to similar X-ray spectra obtained for the best AIGaAs—GaAs superlattices. Fig. 3 shows plots of Hall mobility and carrier concentration versus temperature for two representative n-type modulation-doped samples. Both samples consist of twenty double-layers of 51.8 A —
J.F. Schetzina et aL
/ Modulation-doped HgCdTe QWstructures
Hg0 15Cd0 85Te: In modifier layers separated by 1100 A thick HgCdTe matrix layers. At the top of fig. 3 data for sample B19B having a matrix layer X-value of 0.26 is shown. In this case, the electron increases from about 3000 2/V.s mobility at room temperature to —4700 cm2/V. cm s at low temperatures. The average electron concentration is 3 x 1018 cm3 and is independent of temperature. A reduction of the In oven temperature leads to a corresponding decrease in the electron concentration as shown at the bottom of fig. 3, where Hall data for sample B21B (x = 0.21) are plotted. In this case, n 3 and, 0 =is1.6 X 1017 cm of ternagain, the electron density independent perature. The electron mobility is 8300 cm2/V. s at 300 K and increases to 30,000 cm2/V.s at low temperatures.
27
to~ Indium Doped HgCdTe
—
B22B —0.18 t235 urn
E
~
~•
........_.
Mobili1~
—..--—
Concentration
a
10’~
to
C C.-)
a
i~o Temperature (K)
10
io~0
io~
1017
Indium Doped HgCdTe
—
B24A 1=2.48~ ~18~
—
—
‘-I
..~.
_
and SLi grown by PAMBE
C lOs
1016
cc 1~
~19
Indium Doped HgCdTe
cc
B19B x = 0.26 = 2.32 sm
‘7
O .—_.___
E ll~ 10
C
100 Temperature
______________
a Mobility Concentration
o .
a
I0~
10
100
b
Concentration
1015 1000
(K)
Fig. 4. Hall mobility and carrier concentration versus temperature for n-type modulation-doped structures B22B (a) and B24A (b).
C....)
Fig. 4 shows Hall data for samples B22B and
1018
1000
Temperature (K)
B24A. Both samples consist of twenty doublelayers of 51.8 A Hg 0 15Cd0 85Te: In modifier layers separated by 1130 A thick HgCdTe matrix layers. For B22B, the electron concentration is 8 X 1016 cm ~ and is nearly independent of ternperature. The electron mobility increases as the temperature decreases andFor reaches of 2/V. s at 20 K. growtha ofvalue sample 56,000 cm B24A, the In dopant oven temperature was further reduced. In this case, the electron concentration at temperatures below 100 K is constant and equal to 8 X i015 cm3. Above 100 K thermal activation of carriers occurs. Note that the low temperamobility of sample B24A exceeds 130,000 —
10
Indium Doped HgCdTe -~
,~
C
E
eq
..?:~.
1017
it4
=
B21B = 0.21 t=2.25jsm —0—
cc 7.? C
L)
Mobilily
ture
—
10~
10
100
Temperature
(K)
1000
Fig. 3. Hall mobility and carrier concentration versus temperature for n-type modulation-doped structures B19B (a) and B21B ~.
cm2/V. s. Figs. 5 and 6 show plots of Hall mobility and carrier concentration versus temperature for four representative p-type modulation-doped samples
28
J.F. Schetzina et al.
/ Modulation-doped HgCdTe QW structures and
having layer thicknesses similar to the n-type samples discussed above, except the 50 A modifier —
layers are doped with As to produce p-type modulation-doped samples. At the top of fig. 5, Hall data for a p-type modulation-doped HgCdTe sample (B12A) with matrix layers having an x-value of 0.21 are shown. Note that the Hall coefficient changes sign at T 195 K. Below about T 120 K, both the hole mobility (~, 325 cm2/V. s) and average concentration (Po 5 holes/ cm3) are constant. At the bottom of fig. 5, Hall effect data are shown for p-type sample (B15B) for which x = 0.26. In this case, the Hall coefficient becomes positive at T = 90 K. At low temperatures, the hole mobility ~ 100 cm2/V. s is constant. In fig. 6, corresponding Hall data are —
—
—
—
X
~
= 0.21 B12A I = 2.45 jim
~
f
27.?
“~
o_oon=n=ieIoscu~ -.
?
dPOIO=1t52
I
E
‘°
1010
2
cc
I1.tYPe
:2a
p-type ~
a ~
101 ~
1017
Mobility Concentration
~—0-__.
~
C cc C
a
100
.2
a C~)
Mobility
a
Concentration
10
100
1015
1000
Temperature (K)
IO~
Arsenic Doped HgCdTe
010
(Minority Carrier Transport)
a .2
118A x = 0.85 0.16 Barrie Wells t = 2.3 u~
ID
a
1016
~
C
a
~E Mobility Concentration
0 ..
~oI5
10
100
1000
Temperature (K)
.~
io
E
io’7 ~ C cc 1016 7.? ~
.~ a
C
io2
~
7.?
10
~
10
,-
_~
10
‘-4
B16A x = 0.19 = 2.5 urn
1016
Arsenic bo~edHgCdTe
1020
Ar~e~i~oj,ed HgCdTe
lo~ ~
—
10
SLo grown by PAMBE
Fig. 6. Hall mobility and carrier concentration versus temperature for p-type modulation-doped structures B16A (a) and B8A (b).
1016
10
100
1000
Temperature (K)
1O~
lois
shown for samples B16A (x = 0.19) and B8A (x = 0.16). For B16A, the Hall coefficient becomes positive at 50 K and the low temperature hole concentration is 1.2 x 1017 holes/cm3. The hole mobility is 300 cm2/V. s at 20 K. For sample B8A, the Hall coefficient is negative over the —
Arsenic Doped HgCdTe
—
~.
io~
x
= =
2
~
.~.
0.26 2.35 urn
t018
B15~ io~
10
p-type
:2a
2
C cc
io~
lOll Mobility —~ Concentration 0
100
10
to
a a
b 1015
100 Temperature (K)
1000
Fig. 5. Hall mobility and earner concentration versus temperalure for p-type modulation doped structures B12A (a) and B15B (b).
entiresample temperature range (20~300K), though this is doped with As. This iseven because of the x-value of the matrix layers (x = 0.16) for which the electron mobility ~ is exceedingly large. For B8A, ~ = 380,000 cm2/V s at 50 K. The electron concentration increases monotomically with decreasing temperature and reaches a value of 4 x 1017 electrons/cm3 at 20 K. Note that these date reflect the properties of minority electrons in near-zero band gap (x = 0.16) p-type modulationdoped HgCdTe.
J.F. Schetzina et al. / Modulation-doped HgCdTe
3.2. Modulation-doped HgCdTe superlattices
layer thickness is L2 = 90.7 A. For this superlattice, the Hall coefficient is negative at room temperature due to thermal excitation of carriers, since the electron mobility in HgCdTe is much larger than the hole mobility. The Hall coefficient becomes positive at 180 K and at lower temperatures the holes manifest their properties. For B75A, it is seen that the hole mobility increases monotonically as theoftemperature decreases reaches a maximum ji.~,= 850 cm2/V. s atand 30 K, the lowest temperature measured. This is an excellent mobility value for Hg 0 78Cd022Te, cially since the quantum confinement shiftsespethe band gap of B75B at 4.2 K to nearly 8 tom, corresponding to a bulk alloy of x 0.26. Note, in addition, from fig. 8b the very bright and extremely narrow (FWHM = 10 meY) PL peak that is observed at 4.2 K. For this superlattice, the low temperature hole concentration is not constant, as would be expected for a modulation-doped structure, suggesting that some of the dopant ions may have diffused across the layer interfaces to provide centers for hole freeze-out at low temperatures. Alternatively, hole freeze-out may be due to other
thicknesses are estimates, rounded to the nearest monolayer, that were obtained from the total thickness of the superlattice as measured with a Dektak surface profilometer, the measured X-ray diffraction satellite spacing of the superlattice, and from an analysis of its optical absorption and photoluminescence spectra. Note from the figure that B77A is n-type and exhibits an essentially flat 3 at terncarrier concentration of 3.5 x 1016 cm peratures below 200 K. The electron mobility is 7 x i0~cm2/V. s at 300 K and increases mono4 cm2/V s at 30 K. The phototonically to 3 Xspectrum i0 luminescence obtained for B77A is shown in fig 7b. It consists of a main PL peak centered at 118 meV (10.5 ~ttm) having a fullwidth-at-half-maximum (FWHM) of 30 meV. A high energy shoulder is also seen at 130 meV. We attribute the observed PL peaks as being associated with transitions between the allowed quantum states of the superlattice, with the pmcipal peak corresponding to the energy released by an electron in going from the n = 1 conduction band state to the heavy hole ground state (1H
—
—
—
I0~
Wavelength (i.cm) 10876 5 4 ________________________________ ~iiu~atii~~
11
Modulation-Doped HgCdTe (B77A) n-type
200 Double-Layers In-Doped Barriers = 0.22 Wells
29
transition), which corresponds to the band gap energy of the superlattice (114 meY at 4.2 K). Fig. 8a shows plots of Hall mobility and carrier concentration versus temperature for a p-type modulation-doped HgCdTe superlattice (B75A) consisting of 200 double layers in which the barrier layer thickness is Lb = 51.8 A and the well
Fig. 7a shows plots of Hall mobility and carrier concentration versus temperature for a representative n-type modulation-doped HgCdTe superlattice (B77A) consisting of 200 double layers in which the barrier layer thickness is Lb = 51.8 A and the well layer thickness is L2 = 103.7 A. These
lot
Q Wstructures and SLe grown by PAMBE
-. .
.~
2
~
(B77A) n-type
—
~
200 Doobte-Layers In-Doped Barriers = 0.22 Wells
~ 2 6.
FWI0M
=
30 rneV
0)
L
A
0
=
103.7
Lb
=
51.8 A
L5 = 103.7 A Lb=St.0A
0) ° —t-~—
Mobihiy coocosieailoO
2
_______________________________ 10~
10
100
16 10
1000
0 .2
100
~
200
300
Temperature (K) Photon Energy (meW) Fig. 7. (a) Hall mobility and earner concentration versus temperature for a representative n-type modulation-doped superlattice B77A; (b) photolummescence spectrum of B77A at 4.2 K.
30
J.F. Schetzina et aL
/ Modulation-doped
HgCdTe QW structures and SLt grown by PAMBE Wavelength (~.cm)
4
tots
10
~
Modulation-Doped HgCdTe
to ~
(B75A)
200 Double-Layers
p-type
As-Doped Barriers
t
I
t0~
2
—s-
L~= 90.7
~
A
x
102
=
‘7 2
P
~ ~
to
t
a too
=
10 meV
T=4.2K
Q
a a C.-)
Temperature (K)
4 .
200 Double-Layers As-Doped Barriers
FWIOM
—
0.22 Wells
10
(B7SA) p-type
.~ ~
°nJ~
Mobil4y Conontraoon
5
tO~~ 0.?
a 0
6 .
~
L~=5l.8A
8 7
Modulation-Doped HgCdTe
54
L=90.7A a 0
L=5t.1A 5 = 0.22 Wells
2
1016
~
b
0
1000
200
300
Photon Energy (meV) Fig. 8. (a) Hall mobility and carrier concentration versus temperature for a p-type modulation-doped superlattice B75A; (b) photoluminescence from B75A at 4.2 K.
impurities (such as Hg deficiencies) in the quantum well layers. It would be premature to speculate further on this point at this time, since only a few superlattices of this type have been grown to date. Fig. 9 shows Hall effect and photoluminescence data for a p-type modulation-doped superlattice B73A for which L~= Lb = 51.8 A and x = 0.22 for the Hg 1_5Cd~Te well layers. Note that the quantum confinement associated with small L~is manifested by the large energy shift in the PL peak at 4.2 K to 220 meV (5.6 ,.tm), which conesponds to the band gap of an equivalent bulk Hg1 _5Cd5Te alloy having x = 0.31. The hole mobility increases with decreasing temperature, as
shown in fig. 3a, and reaches a value of = 200 cm2/V. s at 30 K, which is quite reasonable for an equivalent alloy of x = 0.31. 3.3. Modulation-doped HgCdTe heterostructures A schematic diagram of the HgCdTe heterostructures grown by photoassisted MBE to study dopant setback effects is shown in fig. lOa. These multilayered structures consist of a 200 A doping layer of Hg03Cd07Te : In onto which is deposited an undoped Hg0 3Cd07Te spacer layer followed by a 2000 A layer of Hg078Cd022Te. The spacer layer thickness was varied from 50 to 420 A in the various MBE growth runs. As shown in fig. —
Wavelength (tim) 3
___________________
Modulation-Doped HgCdTe
(B73A) p-type
~
102
)
200 Double-Layers AS-Doped Barriers
-~-
‘7
a
~
to
E~
1017
~
t
~
~
u~at~~5
(B73A) p-type
4
200 Double-Layers As-Doped Barriers
_______________
0
Temperature (K)
10876
100
200
300
Photon Energy (rneV)
Fig. 9. (a) Hall effect and (b) photoluminescence data for a p-type modulation-doped superlattice B73A for which L2 = Lb Note the shift of the PL peak to 225 meV (5.5 ~sm)due to quantum confinement.
= 51.8
A.
fE Schetzina et al.
/ Modulation-doped HgCdTe QWstructures 10~
HETEROSTRtJCTURE
Modulation-Doped HgCdTe 200
a
2000AHgCdTe(x=O.22) Spacer _200 (100) A CdZnTe HgCdTe:In Substrate (x 0.7) 0)0.7) _____________________ Layer (x =
2a .2
~0)2 e-t
102 ~ too
10
and SLs grown by PAMBE
A In.Doped Barrier
31
1020
(B62B)
~L
420 Spacer xt=2200A =A0.22 — p-type Mobii~ La r Coomo~os
100
Temperature (K)
10 .2 ~~t8 C cc C U b 1017 a ._._._._. 1016 C 1000
Fig. 10. (a) Schematic drawing of HgCdTe heterostructure. 2D electron gas forms at the Hg
0 78Cd022Te—spacer layer interface due to the charge transfer associated with the modulation doping process; (b) Hall effect data for heterostructure B62B.
lOb, for a spacer layer thickness of 420 A, electron transfer from the Hg03Cd07Te: In dopant barrier layer does not occur sufficiently to dope the active to4
a cc acc a C
region n-type. Rather, p-type conduction occurs at low temperatures as seen from the Hall data shown in the figure. In addition, measured hole mo2/V. s isthe quite low. These features suggest bility p~, 40 to crnus that the thick 420 A spacer layer of Hg 03Cd07Te may be the problem. It is difficult to grow high x-value HgCdTe at low substrate temperatures, particularly when the layer thickness exceeds 100 A. For heterostructure B62B, it appears that the poor quality of the spacer layer epitaxy inhibits modulation doping and creates a defective low-mobility channel at the interface between the spacer layer and the
a
Hg0 78Cd 022Te active layer. For thinner spacer layers this is not the case. In fig. 11, Hall effect data are shown for heterostruc-
—
Modulation-Doped HgCdTe (B63B)
C
A A
200 = 100 .2a x =
In-Doped Barrier Spacer Layer
022 2000
A
-—Se-—
Mobilily
_________________________ iI~
10
100
Temperature (K)
1016
1000
loll
Modulation-Doped HgCdTe (B64A) 200 50 AASpacer In-Doped Layer Barrier C
.~.l0~
~o17
.2C 6.
.2a
cc
x=0.21 I
=
2050
A °
—a— 1o~
10
a C.)
Mobility co~siios 1016
100
—
1000
Temperature (K) b B63B Fig. 11. Hall effect data for HgCdTe heterostructures (a) and (b) B64A.
tures B63B and B64A, having undoped Hg03 Cd0 7Te spacer layer thicknesses of 100 A and 50 A, respectively. For both samples, n-type conducat lowcarrier temperatures. The election is observed3 with concentrations of (3—5)mobility tron x 10~~ cm is fairly independent of temperature for both heterostructures, similar to the behavior reported previously [5] and summarized above for multibarrier modulation-doped HgCdTe alloys. For the heterostructure having a 100 A spacer layer (B63B) the electron mobility /~n= 4500 cm2 1’V~s at low temperatures, whereas the heterostructure having a 50 A spacer layer 2/V~s. exhibits These an electron mobility ~t0 = 9000 preliminary results of suggest spacercm layers of 100
32
.J.F. Schetzina et aL
/ Modulation-doped HgCdTe
A must be employed if effective charge transfer is to be realized.
Acknowledgements
The authors wish to acknowledge J. Matthews, M. Bennet and A. Mohan for their assistance With substrate preparation. We have had many mutually-beneficial discussions with T.H. Myers of General Electric Company (Syracuse, NY) concerning various aspects of photoassisted MBE and modulation-doping of Hg-based structures. Dr. J. Schulman of Hughes Research kindly provided us with computer codes for determining superlattice band gaps based on a tight-binding theoretical calculation. Dr. M. Sen of Santa Barbara Research Center (SBRC) provided us with some of the CdZnTe substrates used in this work. Substrate development at SBRC was supported by NRL contract N00014-87-C-2501. Work at NCSU was supported by NSF grant DMR-88-13525 and NRL contract N0001489J2024.
QWstructures and Sis grown by PAMBE
References [1] M. Boukerche, P.S. Wijewamasuniya, S. Sivananthan, 1K. Sou, Y.J. Kim, K.K. Mahavadi and J.P. Faurie, J. Vacuum Sci. Technol. A6 (1988) 2830. [2] ML. Wroge, D.J. Peterman, B.J. Moms, D.J. Leopold, J.G. Broerman and BJ. Feldman, J. Vacuum Sci. Technol. A6 (1988) 2826. [3] J.W. Han, S. Hwang, Y. Lansari, R.L. Harper, Z. Yang, NC. Giles, J.W. Cook, Jr., S. Sen and J.F. Schetzina, Appl. Phys. Letters 54 (1989) 63. [4] R.N. Bicknell, NC. Giles and J.F. Schetzina, Appi. Phys. Letters 48 (1986) 1095. [5] W. Han, S. Hwang, Y. Lansari, R.L. Harper, Z. Yang, NC. Giles, J.W. Cook, Jr., S. Sen and J.F. Schetzina, J. Vacuum Sci. Technol. A7 (1989) 305. [6] L. Esaki and R. Tsu, IBM Research Note RC-2418 (1969). [7] R. Dingle, H.L. Stormer, A.C. Gossard and W. Wiegmann, Appi. Phys. Letters 37 (1978) 805. [8] E.H.C. Parker, Ed., The Technology and Physics of Molecular Beam Epitaxy (Plenum, New York, 1985). [9] R.N. Bicknell, NC. Giles and J.F. Schetaina, Appl. Phys. Letters 50 (1987) 691. [10] K.A. Harris, S. Hwang, D.K. Blanks, J.W. Cook, Jr., N. Otsuka and J.F. Schetzina, J. Vacuum Sci. Technol. A4 (1986) 2061. [11] RN. Bicknell, N.C. Giles, J.F. Schetzina and C. Hitaman, J. Vacuum Sci. Technol. AS (1987) 3059.
23
MODULATION-DOPED HgCdTe QUANTUM WELL STRUCTURES AND SUPERLATHCES GROWN BY PHOTOASSISTED MOLECULAR BEAM EPITAXY J.F. SCHETZINA, J.W. HAN, Y. LANSARJ, N.C. GILES, Z. YANG, S. HWANG and J.W. COOK, Jr. Department of Physics, North Carolina State University, Raleigh, North-Carolina 27695-8202, USA
and N. OTSUKA School of Materials Engineering Purdue University, West Lafayette, Indiana 47907, USA
We report the results of the first systematic study of modulation doping in HgCdTe. The doped multilayers were grown by means of photoassisted molecular beam epitaxy, in which the substrate is illuminated during the film growth process to enhance dopant activation. Indium was used as the n-type dopant and arsenic was used as the p-type dopant. More than forty modulation-doped samples were prepared and studied.
1. Introduction Precise control of the electrical properties of HgCdTe through the addition of substitutional impurities is an essential requirement for the
At North Carolina State University (NCSU), we have successfully circumvented this fundamental problem by employing modulation doping techniques to produce stable p-type alloys of Hg1 _5Cd~Te(x = 0.18—0.26) [3] by means of pho-
fabrication of infrared detectors based on in-situ grown p—n junctions or p-on-n heterojunctions, the most important area of application for HgCdTe at present. However, growth of stable p-type HgCdTe by MBE (a mercury-deficient growth process) has not been accomplished to date. Boukerche et al. [1] have recently reviewed the problems associated with the use of both Group I elements (Li, Ag, Na) and Group V elements (Sb, As) as potential p-type dopants in HgCdTe. It appears at present that none of these elements are suitable for use as p-type dopants in MBE-grown HgCdTe. The Group I elements lack stability they are fast diffusers in HgCdTe. Thus, one must question the long term viability of HgCdTe homojunctions of heterojunctions which contain these dopants [2]. The Group V elements appear to be amphoteric and often act to produce n-type material due, perhaps, to the Hg-deficient growth conditions which prevail during epitaxial growth of HgCdTe by MBE [1].
toassisted MBE [4]. n-Type modulation-doped samples of Hg1_~Cd~Te (x = 0.18—0.26) have also been successfully prepared [5]. In our initial studies, 8-doped Hg015Cd085Te barrier layers of thickness Lb 50 A were alternated with Hg1....5Cd5Te (x = 0.18—0.26) matrix layers of thickness L~ 1000 A. This produced a new material a quantum alloy of HgCdTe the properties of which are controllable on an atomic basis. Recently, we have completed the growth of new modulation-doped quantum well structures and superlattices of HgCdTe by means of photoassisted MBE technique. In this paper, details of the MBE growth experiments are given along with a discussion of the structural, electrical and optical properties that these interesting new HgCdTe quantum structures exhibit. Modulation doping involves the transfer of carriers (electrons or holes) from a substitutionally doped layer (modifier layer) to an adjacent material having a smaller band gap (matrix layer).
—
0022-0248/90/$03.50 © 1990
—
—
—
—
Elsevier Science Publishers By. (North-Holland)
—
24
J.F. Schetzina et al.
/ Modulation-doped HgCdTe QWstructures and SLa grown by PAMBE
In 1969, in their pioneering work on superlattices, Esaki and Tsu [6] proposed a selectively~doped heterojunction structure for enhanced carrier transport parallel to the interface. Independently in 1978, Dingle et al. [7], using n-type AIGaAs— GaAs heterostructures, first demonstrated that such enhanced carrier transport does, indeed, occur and first evoked the name “modulation doping”. Since then, modulation-doped structures composed of Ill—V semiconductor layers have been studied in detail, and the properties that these novel structures possess have subsequently been exploited in many planar device applications [8]. In 1987, n-type modulation-doped CdMnTe: In—CdTe quantum well structures were successfully prepared for the first time by photoassisted MBE at NCSU [9].
2. Experimental details The modulation-doped HgCdTe samples were grown in a Hg-compatible MBE system designed and built at NCSU. A description of this MBE system together with the techniques that have been developed to grow HgCdTe by conventional MBE are given in an earlier publication [10]. The Hg-MBE system was modified for photoassisted MBE growth experiments by replacing one of the MBE ovens on the main source flange with a UHV Pyrex window. The broad-band output (458—514 nm) from a Spectra-Physics model 201605 argon ion laser was beam-expanded and passed through the window to allow uniform illumination of the substrate during film growth. The laser power densities employedsurface ranged fromsubstrate 40—100 2 at the substrate so that mW/cm effects due to the incident laser beam are heating negligible. In our most recent work, p-type modulationdoped HgCdTe superlattices were grown by alternately depositing Hg 0 3Cd07Te barrier layers that were heavily doped with arsenic with undoped Hg078Cd022Te layers. In this way, superlattices consisting of 200 double layers were prepared. The thickness of the Hg0 3Cd07Te: As doping layers was Lb 50 A in each of the superlattices, with the As dopant uniformly distributed (no setback). —
For the small-band-gap well layers, Hg0 78Cd022Te of thickness L2 = 50—104 A was used in the various superlattice growth experiments. To prepare n-type modulation-doped HgCdTe superlattices, the wide-band-gap Hg03Cd07Te barrier layers were doped with indium rather than arsenic. pType and n-type modulation-doped superlattices consisting of 200 double layers of Hg0 3Cd07Te— HgTe have also been prepared and studied. Preliminary growth experiments of n-type modulation-doped heterostructures of HgCdTe which contain an undoped spacer layer have also been completed. These multilayered structures consist of a 200 A doping layer of Hg03Cd07Te: In onto which is deposited an undoped Hg0 3Cd07Te spacer layer followed by a 2000 A layer of Hg078Cd022Te. The ,spacer layer thickness was varied from 50—420 A in the various MBE growth runs. All of the modulation-doped HgCdTe samples were grown using lattice-matched (100) Cd096 Zn0~Tesubstrates. The substrates were polished and etched using standard techniques [11]. Immediately prior to film growth, the substrates were preheated at 300°Cfor 10 mm in the MBE chamber to drive off residual impurities and to insure a stoichiometric growth surface. A substrate temperature of 170°Cwas used throughout the HgCdTe film growth experiments. The structural perfection of the MBE-grown multilayers was assessed by means of doublecrystal X-ray rocking curve experiments performed using a Blake double-crystal diffractometer equipped with a Philips X-ray generator (Cu Ka X-rays). This instrument has a rocking curve resolution of 1 cross-section arc sec. Vertical transmission electron microscopy (TEM) micrographs were obtained at Purdue University using a JEOL model JEM 2000-EX microscope. Cross-sectional specimens for TEM studies were prepared by mechanical —
grinding followed by ion milling. During the ion milling operation, the specimens were cooled with liquid nitrogen to avoid intermixing of layers. Electrical characterization experiments consisted of Van der Pauw—Hall effect measurements performed on each modulation-doped sample over the temperature range 20—300 K.
J.F. Schetzina et aL
/ Modulation-doped HgCdTe QWstructures and
Infrared photoluminescence (PL) measurements at 4.2 K were completed on selected sampies using a Nicolet 60 SXR Fourier-transform infrared spectrometer (FTIR). In these experiments, the PL signal was excited using the 1.06 ~smemission from a Nd YAG laser focussed to a spot on the sample surface to2.give a power density of approximately 200 W/cm 3. Results and discussion 3.1. Modulation-doped HgCdTe with periodic 8doped barrier layers
p., .
The periodic nature of the modulation-doped HgCdTe samples containing 6-doped Hg 015
SL~grown by PAMBE
25
Cd 085Te barrier layers is clearly manifested by the vertical cross-section TEM micrographs obtained. Fig. la shows a cross-sectional bright field image of sample B15B. As seen from the figure, this sample consists of a matrix of Hg0 74Cd026Te in which twenty 6-doped Hg015Cd085Te layers of thickness Lb A are located periodically and separated by = 51.8 1130 A of the matrix material. The entire modulation-doped structure is shown in cross-section. Note the high structural quality of the epitaxial layers and their high degree of regularity. This is illustrated in more detail in the high resolution dark field image of B15B shown in fig. lb. The structural perfection of the modulationdoped samples is also illustrated by the X-ray —
diffraction spectra which they exhibit. This is
Q2J~ Fig. 1. (a) Bright field vertical cross-section TEM image of p-type modulation-doped structure B15B. The CdZnTe substrate is to the lower right. Twenty 51.8 A thick layers of Hg0 15Cd055Te doped with As appear as a series of parallel bright lines in the multilayer. Note that the superlattice is essentially dislocation-free. (b) High resolution dark field image of B15B showing seven 51.8 A thick layers of Hg0 15Cd035Te doped with As (bright parallel lines). Note the interfacial sharpness of this p-type modulation-doped structure.
26
J.F. Schetzina et aL o
/ Modulation-doped HgCdTe QW structures and SLs grown by PAMBE 0
..—.
Hg 1_~Cd~Te—Hg 1~Cd~TeSuperlattice — -~
Sample B15B T = + 170 L5 = °C 1180
~
Hg1_~Cd~Te—Hg1_~Cd~Te Superlattice
A
‘~
SBRC CdZnTe Substrate
A
________________________________ .
~ 0
Rocking
Sample B2OA T5=+ L5 170°C = 860
r
A
SBRCCdZnTe
Double crystal X ray
p 2
~
T
Ze, —~
o ‘-‘
_7~6 — —1000
__
+
+4
I
0 ANGLE (arc sec)
9
I
I
•
+5
I
0
1000
-,
51
UA
Sample B15B 170°C
+3
___________________ I
—1000
0 ANGLE (arc see)
2
I
I
+4 I
1000
I
•
Hg1 _~Cd~Te_Hg1_~Cd~Te Superlattice
Hg1~Cd~Te—Hg1_~Cd~Te Superlattice
~ a
2
Sample + L5 B2OA = 860 T5= 170°C
.
~
A
SL Principal Peak
>,
A
Double L5 c~t~ X—ray = 1180
~ ~
/U\ ~SL 30Principal arc sec Peak =
z ~
Rocking Curve
mm ~
Beam Cross Section: ~
_____
o —400
1mm x 1
—200
~M~400~
I-
Substrate \\~3BR~CdZTe FWHM(400) = 20 arc sec
Beam Cross Double Rocking crystal Curve Section: x — ray
5-
x im 0
2 arc sec ~- SBRC CdZnTe Substrate FWHM (400)
‘I
0 200 400 —400 —200 0 ANGLE (arc sec) ANGLE (arc sec) Fig. 2. Double-crystal X-ray diffraction rocking curves for samples B12B and B2OA.
shown in fig. 2 for samples B15B and B2OA. At the left is shown the X-ray diffraction rocking curve of B15B (consisting of 51.8 A modifier layers separated by 1130 A thick Hg0 74Cd026Te matrix layers). Since the modifier layers are placed periodically, the sample is thus a superlattice. This is reflected by the occurrence of X-ray satellite peaks. Seven orders of satellites are shown in spectrum of B15B, attesting to the excellent structural quality that has been achieved. Note that the spectrum was obtained using a double-crystal diffractometer with the detector full open (rocking curve configuration) and that the FWHM of the —
16 arc sec 200 400 =
main superlattice peak and each of the sateffites is 30 arc sec. At the right in fig. 2, corresponding double-crystal rocking curves for an n-type modulation-doped sample (B2OA) are shown. Again, multiple orders of satellites are observed with the SL principal peak having a FWHM = 22 arc sec. These spectra are comparable in quality to similar X-ray spectra obtained for the best AIGaAs—GaAs superlattices. Fig. 3 shows plots of Hall mobility and carrier concentration versus temperature for two representative n-type modulation-doped samples. Both samples consist of twenty double-layers of 51.8 A —
J.F. Schetzina et aL
/ Modulation-doped HgCdTe QWstructures
Hg0 15Cd0 85Te: In modifier layers separated by 1100 A thick HgCdTe matrix layers. At the top of fig. 3 data for sample B19B having a matrix layer X-value of 0.26 is shown. In this case, the electron increases from about 3000 2/V.s mobility at room temperature to —4700 cm2/V. cm s at low temperatures. The average electron concentration is 3 x 1018 cm3 and is independent of temperature. A reduction of the In oven temperature leads to a corresponding decrease in the electron concentration as shown at the bottom of fig. 3, where Hall data for sample B21B (x = 0.21) are plotted. In this case, n 3 and, 0 =is1.6 X 1017 cm of ternagain, the electron density independent perature. The electron mobility is 8300 cm2/V. s at 300 K and increases to 30,000 cm2/V.s at low temperatures.
27
to~ Indium Doped HgCdTe
—
B22B —0.18 t235 urn
E
~
~•
........_.
Mobili1~
—..--—
Concentration
a
10’~
to
C C.-)
a
i~o Temperature (K)
10
io~0
io~
1017
Indium Doped HgCdTe
—
B24A 1=2.48~ ~18~
—
—
‘-I
..~.
_
and SLi grown by PAMBE
C lOs
1016
cc 1~
~19
Indium Doped HgCdTe
cc
B19B x = 0.26 = 2.32 sm
‘7
O .—_.___
E ll~ 10
C
100 Temperature
______________
a Mobility Concentration
o .
a
I0~
10
100
b
Concentration
1015 1000
(K)
Fig. 4. Hall mobility and carrier concentration versus temperature for n-type modulation-doped structures B22B (a) and B24A (b).
C....)
Fig. 4 shows Hall data for samples B22B and
1018
1000
Temperature (K)
B24A. Both samples consist of twenty doublelayers of 51.8 A Hg 0 15Cd0 85Te: In modifier layers separated by 1130 A thick HgCdTe matrix layers. For B22B, the electron concentration is 8 X 1016 cm ~ and is nearly independent of ternperature. The electron mobility increases as the temperature decreases andFor reaches of 2/V. s at 20 K. growtha ofvalue sample 56,000 cm B24A, the In dopant oven temperature was further reduced. In this case, the electron concentration at temperatures below 100 K is constant and equal to 8 X i015 cm3. Above 100 K thermal activation of carriers occurs. Note that the low temperamobility of sample B24A exceeds 130,000 —
10
Indium Doped HgCdTe -~
,~
C
E
eq
..?:~.
1017
it4
=
B21B = 0.21 t=2.25jsm —0—
cc 7.? C
L)
Mobilily
ture
—
10~
10
100
Temperature
(K)
1000
Fig. 3. Hall mobility and carrier concentration versus temperature for n-type modulation-doped structures B19B (a) and B21B ~.
cm2/V. s. Figs. 5 and 6 show plots of Hall mobility and carrier concentration versus temperature for four representative p-type modulation-doped samples
28
J.F. Schetzina et al.
/ Modulation-doped HgCdTe QW structures and
having layer thicknesses similar to the n-type samples discussed above, except the 50 A modifier —
layers are doped with As to produce p-type modulation-doped samples. At the top of fig. 5, Hall data for a p-type modulation-doped HgCdTe sample (B12A) with matrix layers having an x-value of 0.21 are shown. Note that the Hall coefficient changes sign at T 195 K. Below about T 120 K, both the hole mobility (~, 325 cm2/V. s) and average concentration (Po 5 holes/ cm3) are constant. At the bottom of fig. 5, Hall effect data are shown for p-type sample (B15B) for which x = 0.26. In this case, the Hall coefficient becomes positive at T = 90 K. At low temperatures, the hole mobility ~ 100 cm2/V. s is constant. In fig. 6, corresponding Hall data are —
—
—
—
X
~
= 0.21 B12A I = 2.45 jim
~
f
27.?
“~
o_oon=n=ieIoscu~ -.
?
dPOIO=1t52
I
E
‘°
1010
2
cc
I1.tYPe
:2a
p-type ~
a ~
101 ~
1017
Mobility Concentration
~—0-__.
~
C cc C
a
100
.2
a C~)
Mobility
a
Concentration
10
100
1015
1000
Temperature (K)
IO~
Arsenic Doped HgCdTe
010
(Minority Carrier Transport)
a .2
118A x = 0.85 0.16 Barrie Wells t = 2.3 u~
ID
a
1016
~
C
a
~E Mobility Concentration
0 ..
~oI5
10
100
1000
Temperature (K)
.~
io
E
io’7 ~ C cc 1016 7.? ~
.~ a
C
io2
~
7.?
10
~
10
,-
_~
10
‘-4
B16A x = 0.19 = 2.5 urn
1016
Arsenic bo~edHgCdTe
1020
Ar~e~i~oj,ed HgCdTe
lo~ ~
—
10
SLo grown by PAMBE
Fig. 6. Hall mobility and carrier concentration versus temperature for p-type modulation-doped structures B16A (a) and B8A (b).
1016
10
100
1000
Temperature (K)
1O~
lois
shown for samples B16A (x = 0.19) and B8A (x = 0.16). For B16A, the Hall coefficient becomes positive at 50 K and the low temperature hole concentration is 1.2 x 1017 holes/cm3. The hole mobility is 300 cm2/V. s at 20 K. For sample B8A, the Hall coefficient is negative over the —
Arsenic Doped HgCdTe
—
~.
io~
x
= =
2
~
.~.
0.26 2.35 urn
t018
B15~ io~
10
p-type
:2a
2
C cc
io~
lOll Mobility —~ Concentration 0
100
10
to
a a
b 1015
100 Temperature (K)
1000
Fig. 5. Hall mobility and earner concentration versus temperalure for p-type modulation doped structures B12A (a) and B15B (b).
entiresample temperature range (20~300K), though this is doped with As. This iseven because of the x-value of the matrix layers (x = 0.16) for which the electron mobility ~ is exceedingly large. For B8A, ~ = 380,000 cm2/V s at 50 K. The electron concentration increases monotomically with decreasing temperature and reaches a value of 4 x 1017 electrons/cm3 at 20 K. Note that these date reflect the properties of minority electrons in near-zero band gap (x = 0.16) p-type modulationdoped HgCdTe.
J.F. Schetzina et al. / Modulation-doped HgCdTe
3.2. Modulation-doped HgCdTe superlattices
layer thickness is L2 = 90.7 A. For this superlattice, the Hall coefficient is negative at room temperature due to thermal excitation of carriers, since the electron mobility in HgCdTe is much larger than the hole mobility. The Hall coefficient becomes positive at 180 K and at lower temperatures the holes manifest their properties. For B75A, it is seen that the hole mobility increases monotonically as theoftemperature decreases reaches a maximum ji.~,= 850 cm2/V. s atand 30 K, the lowest temperature measured. This is an excellent mobility value for Hg 0 78Cd022Te, cially since the quantum confinement shiftsespethe band gap of B75B at 4.2 K to nearly 8 tom, corresponding to a bulk alloy of x 0.26. Note, in addition, from fig. 8b the very bright and extremely narrow (FWHM = 10 meY) PL peak that is observed at 4.2 K. For this superlattice, the low temperature hole concentration is not constant, as would be expected for a modulation-doped structure, suggesting that some of the dopant ions may have diffused across the layer interfaces to provide centers for hole freeze-out at low temperatures. Alternatively, hole freeze-out may be due to other
thicknesses are estimates, rounded to the nearest monolayer, that were obtained from the total thickness of the superlattice as measured with a Dektak surface profilometer, the measured X-ray diffraction satellite spacing of the superlattice, and from an analysis of its optical absorption and photoluminescence spectra. Note from the figure that B77A is n-type and exhibits an essentially flat 3 at terncarrier concentration of 3.5 x 1016 cm peratures below 200 K. The electron mobility is 7 x i0~cm2/V. s at 300 K and increases mono4 cm2/V s at 30 K. The phototonically to 3 Xspectrum i0 luminescence obtained for B77A is shown in fig 7b. It consists of a main PL peak centered at 118 meV (10.5 ~ttm) having a fullwidth-at-half-maximum (FWHM) of 30 meV. A high energy shoulder is also seen at 130 meV. We attribute the observed PL peaks as being associated with transitions between the allowed quantum states of the superlattice, with the pmcipal peak corresponding to the energy released by an electron in going from the n = 1 conduction band state to the heavy hole ground state (1H
—
—
—
I0~
Wavelength (i.cm) 10876 5 4 ________________________________ ~iiu~atii~~
11
Modulation-Doped HgCdTe (B77A) n-type
200 Double-Layers In-Doped Barriers = 0.22 Wells
29
transition), which corresponds to the band gap energy of the superlattice (114 meY at 4.2 K). Fig. 8a shows plots of Hall mobility and carrier concentration versus temperature for a p-type modulation-doped HgCdTe superlattice (B75A) consisting of 200 double layers in which the barrier layer thickness is Lb = 51.8 A and the well
Fig. 7a shows plots of Hall mobility and carrier concentration versus temperature for a representative n-type modulation-doped HgCdTe superlattice (B77A) consisting of 200 double layers in which the barrier layer thickness is Lb = 51.8 A and the well layer thickness is L2 = 103.7 A. These
lot
Q Wstructures and SLe grown by PAMBE
-. .
.~
2
~
(B77A) n-type
—
~
200 Doobte-Layers In-Doped Barriers = 0.22 Wells
~ 2 6.
FWI0M
=
30 rneV
0)
L
A
0
=
103.7
Lb
=
51.8 A
L5 = 103.7 A Lb=St.0A
0) ° —t-~—
Mobihiy coocosieailoO
2
_______________________________ 10~
10
100
16 10
1000
0 .2
100
~
200
300
Temperature (K) Photon Energy (meW) Fig. 7. (a) Hall mobility and earner concentration versus temperature for a representative n-type modulation-doped superlattice B77A; (b) photolummescence spectrum of B77A at 4.2 K.
30
J.F. Schetzina et aL
/ Modulation-doped
HgCdTe QW structures and SLt grown by PAMBE Wavelength (~.cm)
4
tots
10
~
Modulation-Doped HgCdTe
to ~
(B75A)
200 Double-Layers
p-type
As-Doped Barriers
t
I
t0~
2
—s-
L~= 90.7
~
A
x
102
=
‘7 2
P
~ ~
to
t
a too
=
10 meV
T=4.2K
Q
a a C.-)
Temperature (K)
4 .
200 Double-Layers As-Doped Barriers
FWIOM
—
0.22 Wells
10
(B7SA) p-type
.~ ~
°nJ~
Mobil4y Conontraoon
5
tO~~ 0.?
a 0
6 .
~
L~=5l.8A
8 7
Modulation-Doped HgCdTe
54
L=90.7A a 0
L=5t.1A 5 = 0.22 Wells
2
1016
~
b
0
1000
200
300
Photon Energy (meV) Fig. 8. (a) Hall mobility and carrier concentration versus temperature for a p-type modulation-doped superlattice B75A; (b) photoluminescence from B75A at 4.2 K.
impurities (such as Hg deficiencies) in the quantum well layers. It would be premature to speculate further on this point at this time, since only a few superlattices of this type have been grown to date. Fig. 9 shows Hall effect and photoluminescence data for a p-type modulation-doped superlattice B73A for which L~= Lb = 51.8 A and x = 0.22 for the Hg 1_5Cd~Te well layers. Note that the quantum confinement associated with small L~is manifested by the large energy shift in the PL peak at 4.2 K to 220 meV (5.6 ,.tm), which conesponds to the band gap of an equivalent bulk Hg1 _5Cd5Te alloy having x = 0.31. The hole mobility increases with decreasing temperature, as
shown in fig. 3a, and reaches a value of = 200 cm2/V. s at 30 K, which is quite reasonable for an equivalent alloy of x = 0.31. 3.3. Modulation-doped HgCdTe heterostructures A schematic diagram of the HgCdTe heterostructures grown by photoassisted MBE to study dopant setback effects is shown in fig. lOa. These multilayered structures consist of a 200 A doping layer of Hg03Cd07Te : In onto which is deposited an undoped Hg0 3Cd07Te spacer layer followed by a 2000 A layer of Hg078Cd022Te. The spacer layer thickness was varied from 50 to 420 A in the various MBE growth runs. As shown in fig. —
Wavelength (tim) 3
___________________
Modulation-Doped HgCdTe
(B73A) p-type
~
102
)
200 Double-Layers AS-Doped Barriers
-~-
‘7
a
~
to
E~
1017
~
t
~
~
u~at~~5
(B73A) p-type
4
200 Double-Layers As-Doped Barriers
_______________
0
Temperature (K)
10876
100
200
300
Photon Energy (rneV)
Fig. 9. (a) Hall effect and (b) photoluminescence data for a p-type modulation-doped superlattice B73A for which L2 = Lb Note the shift of the PL peak to 225 meV (5.5 ~sm)due to quantum confinement.
= 51.8
A.
fE Schetzina et al.
/ Modulation-doped HgCdTe QWstructures 10~
HETEROSTRtJCTURE
Modulation-Doped HgCdTe 200
a
2000AHgCdTe(x=O.22) Spacer _200 (100) A CdZnTe HgCdTe:In Substrate (x 0.7) 0)0.7) _____________________ Layer (x =
2a .2
~0)2 e-t
102 ~ too
10
and SLs grown by PAMBE
A In.Doped Barrier
31
1020
(B62B)
~L
420 Spacer xt=2200A =A0.22 — p-type Mobii~ La r Coomo~os
100
Temperature (K)
10 .2 ~~t8 C cc C U b 1017 a ._._._._. 1016 C 1000
Fig. 10. (a) Schematic drawing of HgCdTe heterostructure. 2D electron gas forms at the Hg
0 78Cd022Te—spacer layer interface due to the charge transfer associated with the modulation doping process; (b) Hall effect data for heterostructure B62B.
lOb, for a spacer layer thickness of 420 A, electron transfer from the Hg03Cd07Te: In dopant barrier layer does not occur sufficiently to dope the active to4
a cc acc a C
region n-type. Rather, p-type conduction occurs at low temperatures as seen from the Hall data shown in the figure. In addition, measured hole mo2/V. s isthe quite low. These features suggest bility p~, 40 to crnus that the thick 420 A spacer layer of Hg 03Cd07Te may be the problem. It is difficult to grow high x-value HgCdTe at low substrate temperatures, particularly when the layer thickness exceeds 100 A. For heterostructure B62B, it appears that the poor quality of the spacer layer epitaxy inhibits modulation doping and creates a defective low-mobility channel at the interface between the spacer layer and the
a
Hg0 78Cd 022Te active layer. For thinner spacer layers this is not the case. In fig. 11, Hall effect data are shown for heterostruc-
—
Modulation-Doped HgCdTe (B63B)
C
A A
200 = 100 .2a x =
In-Doped Barrier Spacer Layer
022 2000
A
-—Se-—
Mobilily
_________________________ iI~
10
100
Temperature (K)
1016
1000
loll
Modulation-Doped HgCdTe (B64A) 200 50 AASpacer In-Doped Layer Barrier C
.~.l0~
~o17
.2C 6.
.2a
cc
x=0.21 I
=
2050
A °
—a— 1o~
10
a C.)
Mobility co~siios 1016
100
—
1000
Temperature (K) b B63B Fig. 11. Hall effect data for HgCdTe heterostructures (a) and (b) B64A.
tures B63B and B64A, having undoped Hg03 Cd0 7Te spacer layer thicknesses of 100 A and 50 A, respectively. For both samples, n-type conducat lowcarrier temperatures. The election is observed3 with concentrations of (3—5)mobility tron x 10~~ cm is fairly independent of temperature for both heterostructures, similar to the behavior reported previously [5] and summarized above for multibarrier modulation-doped HgCdTe alloys. For the heterostructure having a 100 A spacer layer (B63B) the electron mobility /~n= 4500 cm2 1’V~s at low temperatures, whereas the heterostructure having a 50 A spacer layer 2/V~s. exhibits These an electron mobility ~t0 = 9000 preliminary results of suggest spacercm layers of 100
32
.J.F. Schetzina et aL
/ Modulation-doped HgCdTe
A must be employed if effective charge transfer is to be realized.
Acknowledgements
The authors wish to acknowledge J. Matthews, M. Bennet and A. Mohan for their assistance With substrate preparation. We have had many mutually-beneficial discussions with T.H. Myers of General Electric Company (Syracuse, NY) concerning various aspects of photoassisted MBE and modulation-doping of Hg-based structures. Dr. J. Schulman of Hughes Research kindly provided us with computer codes for determining superlattice band gaps based on a tight-binding theoretical calculation. Dr. M. Sen of Santa Barbara Research Center (SBRC) provided us with some of the CdZnTe substrates used in this work. Substrate development at SBRC was supported by NRL contract N00014-87-C-2501. Work at NCSU was supported by NSF grant DMR-88-13525 and NRL contract N0001489J2024.
QWstructures and Sis grown by PAMBE
References [1] M. Boukerche, P.S. Wijewamasuniya, S. Sivananthan, 1K. Sou, Y.J. Kim, K.K. Mahavadi and J.P. Faurie, J. Vacuum Sci. Technol. A6 (1988) 2830. [2] ML. Wroge, D.J. Peterman, B.J. Moms, D.J. Leopold, J.G. Broerman and BJ. Feldman, J. Vacuum Sci. Technol. A6 (1988) 2826. [3] J.W. Han, S. Hwang, Y. Lansari, R.L. Harper, Z. Yang, NC. Giles, J.W. Cook, Jr., S. Sen and J.F. Schetzina, Appl. Phys. Letters 54 (1989) 63. [4] R.N. Bicknell, NC. Giles and J.F. Schetzina, Appi. Phys. Letters 48 (1986) 1095. [5] W. Han, S. Hwang, Y. Lansari, R.L. Harper, Z. Yang, NC. Giles, J.W. Cook, Jr., S. Sen and J.F. Schetzina, J. Vacuum Sci. Technol. A7 (1989) 305. [6] L. Esaki and R. Tsu, IBM Research Note RC-2418 (1969). [7] R. Dingle, H.L. Stormer, A.C. Gossard and W. Wiegmann, Appi. Phys. Letters 37 (1978) 805. [8] E.H.C. Parker, Ed., The Technology and Physics of Molecular Beam Epitaxy (Plenum, New York, 1985). [9] R.N. Bicknell, NC. Giles and J.F. Schetaina, Appl. Phys. Letters 50 (1987) 691. [10] K.A. Harris, S. Hwang, D.K. Blanks, J.W. Cook, Jr., N. Otsuka and J.F. Schetzina, J. Vacuum Sci. Technol. A4 (1986) 2061. [11] RN. Bicknell, N.C. Giles, J.F. Schetzina and C. Hitaman, J. Vacuum Sci. Technol. AS (1987) 3059.