662 Nuclear Instruments and Methods in Physics Research B24/25 (1987) 662-666
North-Holland, Amsterdam
MEASUREMENT OF ABSOLUTE AI CONCENTRATION IN AI,`Ga 1-,,As
D. YAN *, J. Paul FARRELL, P.M.S. LESSER * and F.H. POLLAK *
Department of Physics, Brooklyn College of CUNY, Brooklyn, N Y 11210, USA
T.F. KUECH and D.J. WOLFORD
IBM Thomas J. Watson Research Center, Yorktown Heights, N Y 10598, USA
A technique has been developed to measure the absolute AI concentration x in AI~GaI_~,As. The technique involves
simultaneous measurements of the 27Al(p, v)ZSsi resonant nuclear reaction and Rutherford backscattering (RBS), and comparisons
with an AlAs reference sample. A detailed description of the experimental procedure is given. Samples analyzed in this study were
prepared by LPE and MOVPE growth on GaAs substrates, with epitaxial layer thicknesses in the range 1-3 /.tm and AI
concentrations in the range 0.10 < x < 0.85. Measurements of x with an absolute error < 0.02 were obtained. Factors limiting the
precision obtainable with this technique are discussed.
1. Introduction
Among the newer semiconductor materials currently
under development, AlxGax_xAs/GaAs has dearly
demonstrated its advantageous electrical and optical
properties. Applications include devices such as cw room
temperature lasers [1], quantum well lasers [2], modula-
tion doped high electron mobility transistors [3], and
high efficiency solar cells [4,5], to mention a few. A
number of other semiconductor materials containing
aluminum, including the ternary compounds Al~,Gal_x
Sb, Al:,In l_:~As and Al~,In1_~,P, show promise of being
similarly very useful for device applications. The prop-
erties of wafers and heterojunction devices, made from
these materials, depend sensitively on the alloy com-
position and for this reason it is important to have
precise and reliable methods of measuring the absoIute
aluminum concentration, x, in materials such as A1x
Gal_xAs.
To date, only two primary techniques have been
available for determining alloy composition: chemical
analysis and electron microprobe analysis. The electron
microprobe technique [6,7], when used for analyzing
heterostructures having epilayer thicknesses in excess of
a few thousand hngstr~Sms, gives results with a claimed
relative accuracy of 2-5%. However, this technique is
time consuming at best and somewhat difficult to use
for very thin films. Therefore, secondary analysis tech-
* Also at Physics Department, Graduate School and Univer-
sity Center, City University of New York, New York, NY
10036, USA.
0168-583X/87/$03.50 © Elsevier Science Publishers B.V.
(North-Holland Physics Publishing Division)
niques such as photoluminescence [7], electroreflectance
[8,9], and Raman spectroscopy [10] are commonly
utilized to determine x in Al~,Gal_xAs and related
compounds. Secondary techniques rely on the fact that
optical band gaps or phonon frequencies (or some
quantities dependent on them) are monotonic functions
of x. The exact relationship between these quantities
and x can only be determined experimentally on the
basis of primary measurements of x, and therefore the
reliability of secondary techniques is limited by the
reliability of primary measurements. For example, there
are some significant discrepancies in the available litera-
ture on photoluminescence peak energy versus alumi-
num concentration in AlxGal_,`As [7]. It is important,
therefore, to look for alternative primary techniques for
measuring aluminum concentration, and to compare
results obtained by different methods, in order to
establish reliable calibrations of secondary measure-
ment techniques.
Ion-beam analysis, via Rutherford backscattering
(RBS) and/or nuclear reaction analysis (NRA), pro-
vides another primary technique for measuring com-
positions [11]. In particular, NRA provides a.sensitive,
direct method of measuring the absolute AI concentra-
tion by measuring the y-ray yield from the 27Al(p, y)2SSi
reaction. This reaction has many (well-known) isolated,
narrow resonances at low energies [12], the dominant
one occurring at an incident proton energy of 992 keV.
In the present work, we have employed a combination
of NRA and RBS to obtain precise, reliable measure-
ments of x in AI~,Gal_,`As. The technique is not,
however, without some difficulties and limitations which
will be discussed in this paper.