Effect of interfacial dopant layer on transport properties of high purity InP

Article abstract of DOI:10.1063/1.115639

Temperature dependent Hall measurements were used to investigate the role of unintentional interfacial dopant layers on the electrical properties of high purity InP. Secondary ion mass spectroscopy was used to measure the level of Si contamination present in several samples at the substrate-epilayer interface. We show that the presence of interfacial dopant gives rise to two layer conduction whose temperature dependence mimics the freezeout behavior expected for a deep donor. However, the magnetic field dependence of the Hall data at low temperatures shows the expected behavior for the two layer model, including an order of magnitude variation in the apparent sheet concentration and Hall mobility over the range from 0.1 to 0.6 T at 77 K. We show that the true bulk mobility and carrier concentration can be accurately determined in samples with high interfacial contamination by performing Hall measurements at several magnetic fields at 77 K.

Full text of DOI:10.1063/1.115639

Effect of interfacial dopant layer on transport properties of high purity InP
S. P. Watkins, H. D. Cheung, G. Knight, and G. Kelly
Citation: Applied Physics Letters 68, 1960 (1996); doi: 10.1063/1.115639
View online: http://dx.doi.org/10.1063/1.115639
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/68/14?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
ptype GaSb and Ga0.8In0.2Sb layers grown by metalorganic vapor phase epitaxy using silane as the dopant
source
Appl. Phys. Lett. 69, 3863 (1996); 10.1063/1.117130
Low resistance (1×10−6 Ωcm2) Au/Ge/Pd Ohmic contact to nAl0.5In0.5P
J. Vac. Sci. Technol. B 14, 3244 (1996); 10.1116/1.588814
Growth and electrical characterization of Si deltadoped GaInP by low pressure metalorganic chemical vapor
deposition
J. Appl. Phys. 79, 8054 (1996); 10.1063/1.362359
Dopantinduced lattice dilation in ntype InP homoepitaxial layers
J. Appl. Phys. 79, 6890 (1996); 10.1063/1.361512
Optical study of good quality InGaP/GaAs quantum wells: Influence of the indium content around the lattice
matched composition
Appl. Phys. Lett. 68, 2111 (1996); 10.1063/1.115601
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 129.174.21.5
On: Thu, 18 Dec 2014 22:04:15

Effect of interfacial dopant layer on transport properties of high purity InP
S. P. Watkins and H. D. Cheung
Department of Physics, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada
G. Knight and G. Kelly
Bell Northern Research, Ottawa, Ontario K1Y 4H7, Canada
͑Received 6 November 1995; accepted for publication 29 January 1996͒
Temperature dependent Hall measurements were used to investigate the role of unintentional
interfacial dopant layers on the electrical properties of high purity InP. Secondary ion mass
spectroscopy was used to measure the level of Si contamination present in several samples at the
substrate-epilayer interface. We show that the presence of interfacial dopant gives rise to two layer
conduction whose temperature dependence mimics the freezeout behavior expected for a deep
donor. However, the magnetic field dependence of the Hall data at low temperatures shows the
expected behavior for the two layer model, including an order of magnitude variation in the apparent
sheet concentration and Hall mobility over the range from 0.1 to 0.6 T at 77 K. We show that the
true bulk mobility and carrier concentration can be accurately determined in samples with high
interfacial contamination by performing Hall measurements at several magnetic fields at 77 K.
© 1996 American Institute of Physics. ͓S0003-6951͑96͒03214-4͔
2
n ␮ ϩn ␮ ͒2ϩ ␮ ␮ ͒ n ϩn ͒2B2
Control of residual impurities is a key issue in the fab-
͑
͑
͑
1
1
1
2
2
1
2
2
ntϭ
,
͑1͒
͑2͒
rication of compound semiconductor devices based on GaAs
and InP. In recent years the technique of metalorganic
chemical vapor deposition ͑MOCVD͒ has achieved great
success in reducing impurity incorporation to extremely low
levels. Hall methods are often a primary characterization tool
giving valuable information regarding the cleanliness of the
reactor and the source chemicals. In this letter, we indicate
an important pitfall of performing such Hall measurements at
one value of the magnetic field. We show that the presence
of an interfacial dopant layer at the epilayer-substrate bound-
ary can give rise to anomalous Hall data, which can lead to
several wrong conclusions regarding the electrical properties
of the bulk material. Interfacial dopant layers have serious
detrimental implications for devices, especially if regrowth is
required. Devices which depend upon a high purity buffer
layer, such as high electron mobility transistors ͑HEMTs͒,
field effect transistors ͑FETs͒, and metal-semiconductor-
metal ͑MSM͒ photodetectors are especially sensitive to im-
purity spikes at the epilayer-substrate interface. For example,
impurity spikes in HEMTs or FETs can cause difficulties in
obtaining good pinch-off characteristics resulting in inferior
noise properties. In this letter, we show that a proper analysis
of Hall measurements can be used to assess the degree of
interfacial contamination of epitaxial layers as well as pro-
viding a technique for measuring the true bulk electrical
properties.
The role of two carrier or two layer conduction in inter-
preting Hall measurements has been reviewed by Stradling.1
Intrinsic two layer conduction is commonly observed in nar-
row gap semiconductors such as InAs which suffer from an
accumulation layer at the surface due to the lineup of the
surface levels with respect to the conduction band minimum.
A semiclassical model for the general case of two parallel
conducting sheets is presented in Ref. 1. This model assumes
a Hall factor of unity. In the case of two conducting layers 1
and 2 the apparent sheet concentration and Hall mobilities
are given by
2
2
n ␮ ϩn ␮ ϩ ␮ ␮ ͒ n ϩn ͒B2
2
͑
͑
1
1
2
1
2
2
1
2
n1␮1ϩn2␮
2
␮tϭ
,
nt
where ␮ and ni are the electron mobility and sheet concen-
trations for each of the two layers. An important prediction
of these expressions is a very strong variation in nt and ␮
with magnetic field if there is a large difference between
i
t
␮ and ␮2 . These expressions have been applied with suc-
1
cess to the case of the narrow gap material InAs.2,3 In the
present work we examine a qualitatively similar situation of
high technological importance involving the presence of a
well-characterized Si impurity layer at the substrate-epilayer
interface during growth of high purity epitaxial InP on semi-
insulating InP. In our case one of the layers is the very high
mobility bulk InP epilayer, the other is the unintentional
␦-doped layer containing Si which is often observed by
SIMS measurements at the epilayer-substrate interface.4,5 In
the present work we look at the magnetic field dependence of
the Hall data for such a system, in conjunction with SIMS
measurements, and show how quantitative information re-
garding the properties of both the bulk and interface layers
can be obtained.
Growth of InP was conducted using low pressure
MOCVD at a pressure of 75 Torr and a temperature of
630 °C. Palladium purified hydrogen was used as the ambi-
ent gas, with phosphine and trimethylindium precursors at a
V:III ratio of 230:1. All samples were grown under identical
growth conditions using whole 50 mm diam wafers of ͑100͒
oriented Fe-doped InP, where 4.0 ␮m of undoped InP was
grown at a rate of 1.2 ␮m/h. Substrate preparation was iden-
tical in all cases and included immersion in 5:1:1
H2SO4:H2O2:H2O.
Measurements of the sheet concentrations of Si and O
were determined by secondary ion mass spectroscopy
͑SIMS͒, using a Cameca IMS 4f microanalyzer with a Csϩ
primary beam source. The detected ions were 28Si and 16O
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 129.174.21.5
1960 Appl. Phys. Lett. 68 (14), 1 April 1996 0003-6951/96/68(14)/1960/3/$10.00 © 1996 American Institute of Physics
On: Thu, 18 Dec 2014 22:04:15

TABLE I. Hall data for three different 4 ␮m samples taken at a field of 5000 G together with SIMS concen-
trations of Si and O at the substrate epilayer interface.
Sample
no.
n͑293 K͒
(cm Ϫ2
␮͑293 K͒
n͑77 K͒
(cmϪ2
␮͑77 K͒
SIMS͓SI͔
SIMS͓O͔
)
(103 cm2/V s͒
)
(105 cm2/V s͒
(1012 cm Ϫ2
)
(1012 cm Ϫ2
)
1
2
3
5.7ϫ1010
9.1ϫ1011
1.2ϫ1012
4.9
2.7
2.4
4.8ϫ1010
3.0ϫ1011
3.4ϫ1011
1.6
0.37
0.27
0.40
5.3
7.1
ϳ0
8.6
6.0
negative secondary ions, calibrated using a Si implanted
standard. The oxygen sheet concentration is estimated. Hall
measurements were performed using the van der Pauw
method. A Hall factor of unity was assumed in all calcula-
tions.
with an activation energy of around 31 meV. However, at
temperatures below 10 K, the sheet concentration increases
again, back up to its previous room-temperature value. In
addition, a very strong dependence of the sheet concentration
on magnetic field is observed. The lack of freezeout at low
temperatures is consistent with the high level of interfacial Si
contamination measured by SIMS. The present data can be
easily explained in a two layer picture in which the interfa-
cial Si contributes a high sheet density, low mobility shunt
layer. The low mobility layer suppresses the measured mo-
bility and increases the observed sheet concentration com-
pared with samples without an interfacial layer. The mobility
of the interfacial layer will be essentially temperature inde-
pendent. In contrast, bulk high purity InP has a strong varia-
tion in mobility with a maximum near 50 K. Thus, in the
vicinity of this temperature, the conductivity of the bulk
layer becomes high, resulting in a drop in apparent sheet
concentration due to the fact that a large fraction of the elec-
tron current now flows through the high mobility layer.
The temperature dependent Hall mobility data for
sample 1 are shown in Fig. 2͑a͒. The behavior is very similar
to previous reports of high purity bulk InP. A very small
dependence on magnetic field is observed for this sample as
expected for a single carrier, single layer material. In con-
trast, the Hall mobility data for sample 3 show a very strong
dependence on magnetic field. The discrepancy between the
two fields is largest at temperatures between 10 and 100 K
where the bulk mobility is largest. The mobility data at the
lower magnetic field are also in close agreement with the
mobility data for the layer without interfacial contamination.
As will be shown below, this indicates that the bulk proper-
ties of the two layers are very similar.
Table I shows a brief summary of transport and SIMS
data for three samples reported in this study. All samples
were grown under similar conditions. Hall measurements of
sample 1 show a significantly lower apparent sheet concen-
tration than samples 2 and 3. However, low-temperature
photoluminescence measurements show no appreciable dif-
ference between the samples, with very sharp donor bound
exciton transitions,6 and in addition there is no appreciable
difference between the bulk SIMS data for the three epilay-
ers. The 77 K mobility of samples 2 and 3 are significantly
lower than for sample 1. The lower mobilities and higher
apparent bulk concentrations of samples 2 and 3 correlate
well with a high level of interfacial Si and O contamination
as measured by SIMS. An interesting feature is the large
drop in the apparent bulk concentration as the measurement
temperature is decreased from 293 to 77 K, suggesting at
first sight some form of deep donor freezeout.
Figures 1͑a͒ and 1͑b͒ compare the measured sheet con-
centrations of a sample without interfacial contamination
͑sample 1͒ and with interfacial contamination ͑sample 3͒.
Sample 1 shows the typical freezeout behavior observed for
high purity bulk InP. The observed activation energy of ϳ6.8
meV obtained for the 100 G data agrees reasonably well with
the predicted effective mass donor binding energy of 7.31
meV. The temperature dependent sheet concentration data
for sample 3 are strikingly different, and are shown in Fig.
1͑b͒. The data show a strong drop starting from 293 K,
which is close to exponential in 1/T down to 100 K. This
would appear to indicate some sort of freezeout behavior
The 77 K magnetic field dependence of the carrier con-
centration for samples 1 and 3 is shown in Fig. 3. The high
purity sample without the interfacial layer ͑open symbols͒
FIG. 1. Sheet concentrations vs temperature at two different magnetic fields
for a sample with ͑a͒ low levels of interfacial Si and O and ͑b͒ high levels
of interfacial Si and O.
FIG. 2. Hall mobility vs temperature for samples without interfacial con-
tamination ͑a͒ and with interfacial Si, O contamination ͑b͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 129.174.21.5
Appl. Phys. Lett., Vol. 68, No. 14, 1 April 1996 Watkins et al. 1961
On: Thu, 18 Dec 2014 22:04:15

timates of the bulk mobility and carrier concentration can be
obtained even in samples with high interfacial donor con-
tamination, provided the magnetic field is below 500 G.
Most routine Hall measurements are performed at a single
field of around 2000–5000 G, in which case serious errors
can arise regarding the bulk properties of epitaxial films.
The value for the Si donor density determined from the
Hall analysis is almost an order of magnitude lower than that
obtained by SIMS. Similar discrepancies are commonly ob-
served for activation of bulk donors. In addition, the pres-
ence of high levels of oxygen at the interface may have the
effect of compensating some of the Si donors. Finally, the
semi-insulating substrate certainly should deplete some of
the electron density from the interfacial layer.
We have identified two conditions under which interfa-
cial Si and O contamination of the type reported here occurs.
The two contaminated samples 2 and 3 were grown in the
presence of a very small leak in the inlet manifold to the
reactor. Correction of this problem immediately resulted in
the elimination of the interfacial layer in subsequent samples.
We have also observed similar contamination effects if the
wafers were exposed to ambient air for long periods of time
prior to growth.
In conclusion, we have shown that Hall measurements
are capable of providing detailed information regarding the
presence and sheet density of unintentional interfacial elec-
trical impurities in epitaxial growth. With simple field depen-
dent Hall measurements it is possible to extract both the bulk
and interface carrier densities and mobilities in epitaxial
semiconductors. Without a careful analysis of the field de-
pendence of Hall data, it is possible to arrive at erroneous
conclusions regarding bulk purity levels, including the obser-
vation of fictitious deep donors.
FIG. 3. 77 K Hall mobility and sheet concentration as a function of mag-
netic field for samples 3 ͑solid symbols͒ and 1 ͑open symbols͒. The fits to
sample 3 which has interface contamination, are based on the two layer
conduction model described in the text.
shows a small variation over the field range 100–6000 G
which is likely explained by a small field dependence of the
Hall factor, rH at this temperature. The low field limit of the
Hall mobility is 1.6ϫ105 cm2/V s confirming the excellent
quality of this material. In contrast, sample 3 shows a very
strong field dependence in both the mobility and sheet con-
centration of nearly an order of magnitude ͑closed symbols͒.
This variation is much too large to be accounted for by Hall
factor variations. The solid curve in Fig. 3͑b͒ represents a
least-squared fit to the data based on expression ͑1͒. The bulk
sheet concentration, bulk mobility, and surface mobility were
used as adjustable parameters, while the sheet electron den-
sity from the Si interface layer was estimated from the low
temperature limit of Fig. 1͑b͒ to be 1.0ϫ1012 cmϪ2. Using
this value, we obtained values for the bulk sheet concentra-
tion and mobility of 5.0ϫ1010 cm2 and 1.5ϫ105 cm2/V s,
respectively. The fitted surface sheet mobility of 2.9
ϫ103 cm2/V s is somewhat higher than the low-temperature
value of ϳ1ϫ103 cm 2/V s obtained from Fig. 2͑b͒, and
may indicate a small temperature dependence of this quan-
tity. The bulk transport values for this sample agree very
well with those obtained for sample 1, which has negligible
interface contamination. This is proof that the bulk purity of
the two layers is essentially identical, and that the strikingly
different Hall properties are the result of the interfacial layer
in sample 3. This is also consistent with the lack of a differ-
ence between the PL spectra of the two samples.
This work was supported by the Natural Sciences and
Engineering Research Council of Canada, and the British
Columbia Ministry of Employment and Investment.
1 R. A. Stradling, in Growth and Characterization of Semiconductors, ed-
ited by R. A. Stradling and P. C. Klipstein ͑Adam Hilger, Bristol, 1991͒,
p. 170.
2 H. H. Weider, Appl. Phys. Lett. 25, 206 ͑1974͒.
3 S. P. Watkins, C. A. Tran, H. D. Cheung, G. Soerensen, R. A. Ares, Y.
Lacroix, and M. L. W. Thewalt, J. Electron. Mater. 24, 1583 ͑1995͒.
4
¨
K. Rakennus, J. Likonen, J. Nappi, K. Tappura, H. Asonen, and M. Pessa,
Proceedings of the 5th International Conference on Indium Phosphide
and Related Materials ͑IEEE, Piscataway, NJ, 1993͒, p. 107.
5 N. Pan, J. Carter, J. Elliott, H. Hendriks, S. Brierley, and K. C. Hsieh,
Appl. Phys. Lett. 63, 3029 ͑1993͒.
The analysis presented here indicate that meaningful es-
6 M. L. W. Thewalt ͑private communication͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 129.174.21.5
1962 Appl. Phys. Lett., Vol. 68, No. 14, 1 April 1996 Watkins et al.
On: Thu, 18 Dec 2014 22:04:15