VOLUME 86, NUMBER 12 PHYSICAL REVIEW LETTERS 19 MARCH 2001
Experimental Confirmation of the Predicted Shallow Donor Hydrogen State in Zinc Oxide
S. F. J. Cox,
1,2
E. A. Davis,
3
S. P. Cottrell,
1
P. J. C. King,
1
J. S. Lord,
1
J. M. Gil,
4
H. V. Alberto,
4
R. C. Vilão,
4
J. Piroto Duarte,
4
N. Ayres de Campos,
4
A. Weidinger,
5
R. L. Lichti,
6
and S. J. C. Irvine
7
1
ISIS Facility, Rutherford Appleton Laboratory, Chilton OX11 0QX, United Kingdom
2
Department of Physics and Astronomy, University College London, London WCE 6BT, United Kingdom
3
Department of Physics and Astronomy, University of Leicester, Leicester LEI 7RH, United Kingdom
4
Physics Department, University of Coimbra, P-3004-516 Coimbra, Portugal
5
Hahn-Meitner Institut Berlin, Glienicker Strasse 100, D-14109 Berlin, Germany
6
Physics Department, Texas Tech University, Lubbock, Texas 79409-1051
7
Chemistry Department, University of Wales, Bangor, Gwynedd LL57 2UW, United Kingdom
(Received 10 October 2000)
We confirm the recent prediction that interstitial protium may act as a shallow donor in zinc oxide,
by direct spectroscopic observation of its muonium counterpart. On implantation into ZnO, positive
muons — chemically analogous to protons in this context — form paramagnetic centers below about 40 K.
The muon-electron contact hyperfine interaction, as well as the temperature and activation energy for
ionization, imply a shallow level. Similar results for the cadmium chalcogenides suggest that such
shallow donor states are generic to the II-VI compounds. The donor level depths should serve as a guide
for the electrical activity of interstitial hydrogen.
DOI: 10.1103/PhysRevLett.86.2601 PACS numbers: 61.72.Vv, 71.55.Gs, 76.75.+i
This Letter is in response to the very recent theoretical
result by Van de Walle [1], that interstitial hydrogen can
provide a donor level just below the conduction band in
ZnO and therefore act as a source of n-type conductivity
in this material.
We have detected and characterized the muonium coun-
terpart of this state, adding to the theoretical predictions
precise values of ionization temperature, donor level, and
spin density at the donor site in the paramagnetic (undis-
sociated) state. We also report the systematics of these
parameters among the cadmium chalcogenides, further to
our first observation of such a state in CdS [2]. This com-
bination of experimental and theoretical evidence for a
particular compound, together with data for three others,
establishes the general importance of the shallow donor
state in the II-VI compounds and raises the question of
how widespread such behavior of hydrogen and its iso-
topes might be among other wide-gap semiconductors.
Isolated hydrogen atoms commonly form deep-level
defects in semiconductors, with donor and acceptor states
involving different interstitial locations. The paramagnetic
neutral centers are especially difficult to study spectro-
scopically for two reasons. First, hydrogen defects typi-
cally constitute a negative-U system, so that the neutral
states are only metastable. Calculations for another
wide-gap semiconductor, GaN [3], illustrate this behavior
and may be compared with those for ZnO [1]. Second,
hydrogen is so mobile and reactive that it quickly pairs
with and passivates other defects or dopants, removing
their electrically active levels from the energy gap.
To obtain an atomistic picture of the behavior of
the isolated centers, studies of the analogous states of
muonium are at a considerable advantage [4,5]. In these
experiments, muonium is treated as a light isotope of
hydrogen: Mu m
1
e
2
, with m
Mu
m
H
19. In their
vacuum states, Mu and H have the same Bohr radius and
binding energy to within a fraction of a percent. With
due regard for differences in zero-point energy when
constrained interstitially or chemically bound, muonium
provides a good model for hydrogen, adopting the same
sites and an essentially identical local electronic structure.
It may be detected and characterized readily by the muon
spin rotation (mSR) technique, thanks to the special
properties of muon production and decay, which give
this spectroscopy a particulary high sensitivity per spin.
Following muon implantation in the sample, the evolution
of spin polarization is displayed for up to 5–10 muon
lifetimes (t
m
2.2 ms) — usually before the muonium
encounters or reacts with other defects.
Prior to our own studies of CdS [2], muonium centers in
semiconductors had all been found to have energy levels
deep in the energy gap, with tightly localized electron wave
functions. Thus in Si, for which the correspondence be-
tween Mu and H is best documented [6,7], muonium pro-
vides a donor level at about 200 meV below the conduction
band, defined by its ionization energy at the bond-center
site [6]. The negative-U issue is not fully resolved but the
acceptor level, involving a change to the tetrahedral cage
site, is believed to lie deeper still [7]. A similar metastabil-
ity of neutral muonium and bistability of its charged states
is seen in compound semiconductors such as GaAs [5].
According to Van de Walle’s calculations for H in ZnO
[1], interstitial protons could be sited either antibonding
to oxygen in the wurtzite-type lattice or roughly midway
between adjacent O and Zn atoms. The former may be
imagined as the hydroxyl site which is common in oxides
and the latter is analogous to the bond-center site which
is established in the more covalent semiconductors with
0031-9007 01 86(12) 2601(4)$15.00 © 2001 The American Physical Society 2601