0% found this document useful (0 votes)
29 views

Electron Mobility Enhancement From Coupled Wells in Delta-Doped Gaas

Uploaded by

Ismael Ancona
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
29 views

Electron Mobility Enhancement From Coupled Wells in Delta-Doped Gaas

Uploaded by

Ismael Ancona
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Electron mobility enhancement from

coupled wells in delta-doped GaAs


Cite as: Appl. Phys. Lett. 62, 504 (1993); https://doi.org/10.1063/1.108893
Submitted: 17 August 1992 . Accepted: 23 November 1992 . Published Online: 04 June 1998

X. Zheng, T. K. Carns, K. L. Wang, and B. Wu

ARTICLES YOU MAY BE INTERESTED IN

Delta doping of III–V compound semiconductors: Fundamentals and device applications


Journal of Vacuum Science & Technology A 8, 2980 (1990); https://
doi.org/10.1116/1.576617

Subband and transport calculations in double -type -doped quantum wells in Si


Journal of Applied Physics 99, 033702 (2006); https://doi.org/10.1063/1.2168024

Band parameters for III–V compound semiconductors and their alloys


Journal of Applied Physics 89, 5815 (2001); https://doi.org/10.1063/1.1368156

Appl. Phys. Lett. 62, 504 (1993); https://doi.org/10.1063/1.108893 62, 504

© 1993 American Institute of Physics.


Electron mobility enhancement from coupled wells in delta-doped GaAs
X. Zheng,a) T. K. Carns, K. L. Wang, and B. Wub)
Device Research Laboratory, Electrical Engineering Department, 66-147B Engineering IV;
University of California, Los Angeles, Los Angeles, California 90024-1594
(Received 17 August 1992; accepted for publication 23 November 1992)
It is found that when two Si delta (6) doped wells in GaAs are placed in close proximity to one
another, the electron Hall mobility is enhanced two to five times over that of the single well case.
The temperature dependence of the mobility is also reported. Samples with a variety of spacer
widths have been studied. Our theoretical and experimental data show that the excited subbands
in the double well structure have significant carrier densities located in the undoped region
where Coulombic scattering is reduced, and are thus found to play an important role in the
observed mobility behavior.

Most of the study on electronic transport in &doped Among them, DW200 has the highest Hall mobility and
GaAs has focused on obtaining high electron densities and the smallest temperature dependence. The mobility of
high mobilities by improving doping and material growth DW200 is about two times higher than that of SW at room
techniques. 1-4 Another way of obtaining high conductance temperature and increases to more than five times higher at
is to explore transport using new structures and mecha- 77 K. As the spacer thickness between the coupled S layers
nisms. In this letter, we report a strong electron Hall mo- increases, the mobility drops monotonically, approaching
bility enhancement due to coupling between two (or more) that of SW. This is observed in DW400, where both the
Si &doped wells placed in close proximity to one another. magnitude and temperature dependence of the mobility are
A double well structure with an optimized spacer thickness similar to those of SW, showing characteristics of two-
of 200 A shows a Hall mobility two to five times higher dimensional Coulombic scattering. On the other hand, as
than that of a single well over the temperature region mea- the spacer thickness is decreased to below 200 A, the mo-
sured. Our simulation shows that for appropriately sepa- bility first decreases (for DW 150) and then increases when
rated double S layers, the original two quantum wells are the spacer thickness is about 100 A (for DWlOO). Except
merged to such an extent that a wide well is formed with for DW400, all of the double well samples have a much
two deeper valleys located in the regions where the higher mobility than that of the single well sample. It is
S-doping impurities are placed. interesting to note that the p(r) curves in Fig. 1 exhibit a
The structures used in this study were grown by mo- fanlike feature. From the observed behavior, one can con-
lecular beam epitaxy (MBE) on semi-insulating GaAs clude that the higher the Hall mobility, the weaker the
( 100) substrates at 500 “C!. Planar doping was achieved by temperature dependence. For example, the Hall mobility of
the “growth interruption technique” as follows: close the DW200 has a temperature dependence of T”.05, while
Ga shutter while leaving the As shutter open for 10 s, then those of DW400 and SW are To.4o and T”.5g, respec-
close the As shutter and open the Si shutter for planar tively.
doping.2 In this study, the Si shutter was left open for 60 s For all structures measured, the Hall carrier densities
for all 6 layers grown. This yields a Hall carrier density of are nearly constant with temperature. A typical curve de-
- 5 x 1Or2cmm2. Six different types of samples were grown picting the 2D Hall carrier concentration of DW200 as a
and are designated as SW, DWlOO, DW150, DW200,
DW250, and DW400. SW denotes one single &doped
layer, while DW denotes two coupled S-doped layers with
suffix numerals indicating the thickness of the spacer layer
in A. All of these structures consist of an undoped 1 ,um
buffer layer and an undoped 0.5 pm cap layer. The latter is
used to eliminate the surface depletion into the S layers.5
AuGe/Ni/Au was evaporated onto the sample and the
structures were subsequently sintered in N, ambient at
400 “C for ohmic contact. Electron concentrations and mo-
bilities were measured using the standard van der Pauw
configuration under a magnetic field of 0.54 T.
Shown in Fig. 1 are the Hall mobilities for all six sam-
ples studied over a temperature range from 77 to 420 K.
Temperature (K)
*‘On leave from the Institute of Microelectronics, Tsinghua University, FIG. 1. Hall mobilities for different samples as a function of temperature.
Beijing, China. SW refers to the single well sample and DW refers to the double well
b)Present address: 3M Company, 201-lN-35, 3Mcent, St. Paul, MN samples. The numbers following DW refer to the thickness of the spacer
55144. layer between the wells in A.

504 Appl. Phys. Lett. 62 (5), 1 February 1993 0003-6951/93/050504-03$06.00 @ 1993 American Institute of Physics 504
1500 r--r--f- w-I-12
-P - ......___..___.._.._..~..~~~..,,, 0
?
7 ---
“; 1000 - Ku-) 5 -40
.!T
.5
w”

;i
-80

d‘100
LT..;
O;r~~~~~
w

-120

200 300 400 -40 -20 0 20 40


(4 Z (nm)
Temperature (K)

FIG. 2. Experimentally measured Hall mobility p(r) (solid line), 2D


...I.--
Hall carrier density N(T) (solid line), and the extracted component Hall 0 ._..- .- .._.._
E3
mobilities ~~~(7’) (dotted line) and P~,,(Z’) (dotted line) for DW200. .. ..~..--.- ‘-.-.J ,:“--yq;, T”,,
,.e--,, i E2
Also shown is the carrier density fit by JZq. (2) [dotted line of N( r)]. ... ..._...............-..*..4- %J
El
-40
r=Y P+ f-Y,/

function of temperature, N(T), is shown in Fig. 2. Also -80


shown in Fig. 2 are experimental and simulated mobility
curves, to be discussed later. It is also found that the car-
rier concentration remains nearly constant ( - 1 X lOI
cme2) for temperatures down to 4.2 K, suggesting no car-
rier freezeout.
40 -20 0 20 40
In order to determine the physical mechanism for the .,
Z (nm)
higher electron Hall mobilities in the double well samples,
the quantum-well potential, subband structure, wave func-
tion, and carrier distribution have been calculated for dif- ,““I”“I”“I’~l~~“,~“‘,~~”

ferent spacer thicknesses by solving both the Poisson’s and 0


E3
Schroedinger’s equations simultaneously and self- E2
consistently. In these calculations, the width of the
&doped layer was taken as 20 A. Both rectangular and s -40
E
.” “..-“...“..” . . . . . . . . .._....- k’
El
Gaussian impurity distributions were used and showed lit-
w” -80 J%b-i
tle difference in the results for equivalent doping levels.
Therefore, a rectangular distribution was assumed. The
nonparabolicity of the GaAs conduction band6 and the
many body effect’ were also taken into account in the cal-
culations. Figure 3 shows the results of the calculated po-
tential and probability distributions for samples DW400, -40 -20 0 20 40
DW200, and DWlOO, where the solid curve defines the
z (m)
conduction band edge, E,. When the 6 layers are coupled,
FIG. 3. Calculated probability distributions and potential well for sam-
the confined subbands where the energy is below the cen- ples (a) DW400, (b) DW200, and (c) DWICO. The unlabeled solid
tral barrier (E,,,Eh) will undergo a splitting due to the curve is the conduction band edge, E,(Z), the solid line n(Z) is the
interaction between the two adjacent S layers. However, carrier distribution, and the dotted lines in the potential well indicate the
probability density functions of the different subbands. The impurity dis-
this split is not large and is thus neglected here. It is also tribution is shown by the shaded rectangle at the bottom of each figure.
observed that the excited subbands (E1,53) are global over The Fermi level is also shown by the dashed line near the top of the wells.
the entire well and each symmetric global wave function
has a probability peak located within the undoped spacer
area. Calculations at 0, 77, and 300 K show that because of
subbands transport in &doped GaAs, but did not give fur-
the low density of states, the &doped well in GaAs is
always fully filled and the Fermi level is positioned very ther clarification.2 Here, we propose a simplified two band
close to the top of the well at all temperatures. As a result, model to describe the observed mobility behavior, where
approximately 30% of the total carrier population is lo- we assume that the ground states (Eo,Ei> have an unique
cated in the excited subbands, while the remaining 70% Hall mobility ,ugr, and all the excited states (El,,,.) have
occupy the ground states. This results in multiple subbands another unique Hall mobility pex. The carrier densities of
transport. the ground states and the excited states are designated as
G ilm a n et al. havementionedthe existenceof m u ltiple Ngr and N,., respectively.From the Hall measurementfor-

505 Appl. Phys. Lett., Vol. 62, No. 5, 1 February 1993 Zheng et a/. 505
mulation,* the experimentally measured Hall mobility and culation of the subband structure of DW 150 that the prob-
2D carrier density can be expressed as ability distribution in the spacer region is reduced (as com-
pared to DW200) due to the smaller spacer width.
p= (N,,~~,+Ne~~~)/(N,r~,,+Ne~e,) (1) -As the spacer width is further reduced below 150 A, a
and change takes place in the mobility behavior, as seen for
DWlOO in Fig. 1. Here, we see that the mobility is higher
N= (NgrCLgr+Ne~e~)2/(NgrCL~r+NexCL~~). (2) than that of DW150. This result can be explained from
Shown in Fig. 2 are curves ps,( r) and p,,(T) of sam- Fig. 3(c). In this case, because of the strong coupling be-
ple DW2OO decomposed from the p( 7’) data using formu- tween the wells, the ground states are pushed up to the top
lae ( 1) and (2). The ratio of Ns, to iV,, used for decom- of the central barrier so that they are no longer fully lo-
position was obtained from the calculated carrier calized in the S layers. Instead, they become more global-
population results mentioned previously. The measured 2D like, leading to increased carrier densities in the spacer
carrier density curves of N( T) are shown by the solid line region. This results in a higher asp and thus a higher mo-
and the calculated results by the dashed line. As is evident, bility than that of DW150. The mobility is not as high as
there is reasonable agreement between them. An important DW200 due to the fact that the carriers in the spacer re-
implication for the result shown in Fig. 2 is that the Hall gion are more strongly influenced by the ionized impurity
mobility in the situation of multiple subband transport is centers.
dominated by the carriers populated in the excited (or In conclusion, we have found that closely spaced cou-
global) states, though the population is lower in the excited pled double S-doped wells have much higher Hall mobili-
states than in the ground states. In the excited states, the ties as compared to the single well case. Our simulation
carriers have a much higher probability of appearing in the and theoretical analyses show that this enhancement is
spacer region, in contrast to the ground states where the mainly because of the occupied global states located be-
carriers are located primarily in the impurity layers. Sim- tween the S layers. The occupancy of the global carriers in
ilar to the situation in a modulation doped structure, car- the undoped spacer region is of critical significance in de-
riers appearing in the spacer region will not suffer signifi- termining the mobility. We show that the occupancy of the
cant Coulombic scattering from the impurity centers. global states in the coupled S-doped layers will yield higher
Thus, a higher pex results. Therefore, the mobility will be Hall mobilities due to the fact that they lead to carrier
dictated by the probability distributions of the excited transport which is spatially separated from ionized impu-
states in the spacer region. rity scattering. This mobility enhancement due to wave
Using this conclusion, we can interpret the mobility function coupling has also been observed in Sig and is ex-
behavior of the samples in this study from our self- pected to take place in other semiconductor materials
consistent calculation. Figure 3 provides the results of this which also utilize coupled S layers. The enhanced Hall
calculation as the spacer is reduced from 400 to 100 A. The mobility and the low temperature dependence of the mo-
results for DW400 are shown in Fig. 3 (a). The separation bility provide potential for the coupled S-doped wells to be
between the delta layers is large so that the electron wave used in transport devices. Such a transport layer can be
functions will not couple strongly, leaving very few carriers realized by placing many coupled wells together.
located in the undoped spacer region. Thus, we would ex- Discussions with Dr. R. P. G. Karunasiri on the many
pect DW400 to have a similar mobility behavior to SW. body effect is gratefully acknowledged. This work is sup-
This is found to be the case, as seen in the results shown in ported in part by AR0 (Dr. John Zavada) and AFOSR
Fig. 1. As the spacer width is reduced, the electron wave (Dr. Gerald Witt) .
function interaction between the delta layers becomes in-
creasingly stronger, resulting in more carriers being lo-
cated in the central barrier of the double well structure. ‘P. M. Koenraad, F. A. P. Blom, C. J. G. M. Langerak, M. R. Leys, J.
This will give higher mobilities and reduced mobility tem- A. A. J. Perenboom, J. Singleton, S. J. R. M. Spermon, W. C. ven der
perature dependence, which is observed for both DW250 Vleuten, A. P. J. Voncken, and J. H. Wolter, Semicond. Sci. Technol. 5,
861 (1990).
and DW200. Figure 3 (b) gives the calculation for DW200, ‘G. Gilman, B. Vinter, E. Barbier, and T. Tardella, Appl. Phys. Lett. 52,
showing the probability distribution peaks located in the 972 (1988).
central undoped barrier. As the spacer is reduced to below 3E. F. Schubert, I. E. Cunningham, and W. T. Tsang, Solid State Com-
200 A (for DW 150), the mobility and temperature depen- mun. 63, 591 (1987).
4 W. Ted Masselink, Appl. Phys. Lett. 59, 694 ( 199 1).
dence drop due to the fact that many of the carriers in the ‘A. Chandra, C. E. C. Wood, D. W. Woodard, and L. F. Eastman, Solid ~~
global states are brought in closer proximity to the ionized State Commun. 22, 645 ( 1979).
impurities in the delta layers. This effect is comparable to ‘U. Roessler, Solid State Commun. 49, 943 (1984).
‘B. Jogai, J. Vat. Sci. Technol. B 9, 2473 (1991).
the reduced spacer thickness in modulation doped struc-
*D. K. Schroder, Semiconductor Material and Device Characterization
tures where the mobility is degraded. In addition to the (Wiley, New York, 1990), p. 206.
increased Coulomb scattering, we have found from the cal- 9T. K. Cams, X. Zheng, and K. L. Wang (unpublished).

506 Appl. Phys. Lett., Vol. 62, No. 5, 1 February 1993 Zheng et al. 506

You might also like