
Oxygen vacancies at the origin of pinned moments in oxide interfaces: the example of
tetragonal CuO/SrTiO3
Benjamin Bacq-Labreuil,1, ∗Benjamin Lenz,2, †and Silke Biermann1, 3, 4, 5, ‡
1CPHT, CNRS, Ecole Polytechnique, Institut Polytechnique de Paris, F-91128 Palaiseau, France
2Institut de Min´eralogie, de Physique des Mat´eriaux et de Cosmochimie,
Sorbonne Universit´e, CNRS, MNHN, IRD, 4 Place Jussieu, 75252 Paris, France
3Coll`ege de France, 11 place Marcelin Berthelot, 75005 Paris, France
4Department of Physics, Division of Mathematical Physics,
Lund University, Professorsgatan 1, 22363 Lund, Sweden
5European Theoretical Spectroscopy Facility, Europe
(Dated: October 28, 2022)
Obtaining an accurate theoretical description of the emergent phenomena in oxide heterostruc-
tures is a major challenge. Recently, intriguing paramagnetic spin and pinned orbital moments have
been discovered by x-ray magnetic circular dichro¨ısm measurements at the Cu L2,3-edge of a tetrag-
onal CuO/SrTiO3heterostructure. Using first principles calculations, we propose a scenario that
explains both types of moments, based on the formation of oxygen vacancies in the TiO2interface
layer. We show the emergence of a paramagnetic 2D electron gas hosted in the interface CuO layer.
It is invisible at the Ti L2,3-edge since the valence of the Ti atoms remains unchanged. Strong struc-
tural distortions breaking both the local and global fourfold rotation C4symmetries at the interface
lead to the in-plane pinning of the Cu orbital moment close to the vacancy. Our results, and in
particular the pinning of the orbital moment, may have implications for other systems, especially
monoxide/dioxide interfaces with similar metal-oxygen bond length and weak spin-orbit coupling.
I. INTRODUCTION
Oxide heterostructures are promising hosts for a large
range of applications [1,2], and offer countless possi-
ble configurations allowed by epitaxial thin-film depo-
sition techniques [3]. A particular example are copper
monoxide CuO thin films, which grow in a tetragonal
crystal structure on a SrTiO3(STO) substrate [4,5],
whereas its bulk counterpart displays a low-symmetry
monoclinic structure [6]. Tetragonal CuO (t-CuO) is
made of a staggered stacking of 2D CuO planes along
the caxis. Its tetragonal distortion is characterized by
the apical Cu-O distance being 1.37 times larger than
the basal distance [4,7]. t-CuO is arguably the cuprate
with the simplest crystal structure, making it an ideal
candidate for connecting the low-energy effective models
for cuprates [8–13] to the real materials.
A key step in this direction is the understanding of
the properties of undoped t-CuO. The experimentally
observed tetragonal distortions were reproduced using
density functional theory (DFT) with hybrid function-
als [14], and traced back to Jahn-Teller orbital order-
ing [15,16]. In agreement with resonant inelastic x-ray
scattering (RIXS) [17] and muon spin resonance mea-
surements [18], it was shown that the CuO layers dis-
play an antiferromagnetic stripe order [14–16]. A first
application of strongly-correlated electron techniques to
t-CuO [19] gave an explanation to the sublattice decou-
pling observed in RIXS [17], angle-resolved photoemis-
∗benjamin.bacq-labreuil@polytechnique.edu
†benjamin.lenz@sorbonne-universite.fr
‡silke.biermann@polytechnique.edu
sion spectroscopy [20], as well as scanning tunnelling mi-
croscopy (STM) [5].
This set of experimental and theoretical works would
yield a coherent understanding of undoped t-CuO, if it
was not for the recent finding of an isotropic paramag-
netic spin and pinned orbital moment in t-CuO/STO
samples, by X-ray magnetic circular dichro¨ısm (XMCD)
measurements at the Cu L2,3-edge [18]. The existence of
these moments is puzzling since t-CuO is antiferromag-
netically ordered and does not contain any element with
strong spin-orbit coupling (SOC). Hernandez et al. has
advanced a scenario in Ref. 18, in which t-CuO would
be composed of ferromagnetically ordered CuO layers,
which are antiferromagnetically stacked along the caxis
(out-of-plane). The last CuO layer would be paramag-
netic hence would follow the magnetic field, while an-
other layer would be uncompensated and therefore would
yield the pinned moments. Although this scenario would
qualitatively account for the experimental observations,
it encounters several difficulties: (i) the ferromagnetic or-
dering inside each layer is in contradiction with all previ-
ous experimental [17] and theoretical [14–16,19] findings,
(ii) the uncompensated layer is composed of spin mo-
ments and not orbital ones, whereas the observed pinned
moment is mainly of orbital character, (iii) there is no
reason why one layer should have a paramagnetic behav-
ior, especially if it is ferromagnetically ordered. An alter-
native explanation would reside in the existence of a 2D
electron gas (2DEG) at the t-CuO/STO interface, as was
found in other heterostructures such as BiMnO3/STO,
LaAlO3/STO and γ-Al2O3/STO [21–23]. However, in
these studies the spin moment is observed at the Ti L2,3-
edge, corresponding to the existence of a Ti3+ valence
configuration, contrary to t-CuO/STO for which no Ti3+
arXiv:2210.15084v1 [cond-mat.mtrl-sci] 26 Oct 2022