Guillermina Ramirez San Juan Bloch wave in silicon Optical Lattice.
Electronic structure, optical properties, and lattice ... · 2 Electronic structure, optical...
Transcript of Electronic structure, optical properties, and lattice ... · 2 Electronic structure, optical...
Nano Res
1
Electronic structure, optical properties, and lattice
dynamics in atomically thin Indium Selenide flakes
Juan F. Sánchez-Royo1 (), Guillermo Muñoz-Matutano1,†, Mauro Brotons-Gisbert1, Juan P.
Martínez-Pastor1 (), Alfredo Segura1,2, Andrés Cantarero1, Rafael Mata1, Josep Canet-Ferrer1, Gerard
Tobias3, Enric Canadell3, Jose Marqués-Hueso4, and Brian D. Gerardot4
Nano Res., Just Accepted Manuscript • DOI 10.1007/s12274-014-0516-x
http://www.thenanoresearch.com on June 15, 2014
© Tsinghua University Press 2014
Just Accepted
This is a “Just Accepted” manuscript, which has been examined by the peer-review process and has been
accepted for publication. A “Just Accepted” manuscript is published online shortly after its acceptance,
which is prior to technical editing and formatting and author proofing. Tsinghua University Press (TUP)
provides “Just Accepted” as an optional and free service which allows authors to make their results available
to the research community as soon as possible after acceptance. After a manuscript has been technically
edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP
article. Please note that technical editing may introduce minor changes to the manuscript text and/or
graphics which may affect the content, and all legal disclaimers that apply to the journal pertain. In no event
shall TUP be held responsible for errors or consequences arising from the use of any information contained
in these “Just Accepted” manuscripts. To cite this manuscript please use its Digital Object Identifier (DOI® ),
which is identical for all formats of publication.
Nano Research
DOI 10.1007/s12274-014-0516-x
1
Electronic structure, optical properties, and
lattice dynamics in atomically thin Indium
Selenide flakes
J. F. Sánchez-Royo1,*, G. Muñoz-Matutano1, M.
Brotons-Gisbert1, J.P. Martínez-Pastor1,*, A. Segura1,
A. Cantarero1, R. Mata1, J.Canet-Ferrer1, G. Tobias2,
E. Canadell2, J. Marqués-Hueso3, and B.D. Gerardot3
1Universidad de Valencia, Spain.
2Institut de Ciència de Materials de Barcelona
(ICMAB-CSIC), Spain.
4Heriot-Watt University, UK.
Summary of the work:
We show that quantum size confinement tunes the
dimensionality of the lattice dynamics, optical and
electronic properties of atomically thin InSe flakes
prepared by micromechanical cleavage. Reported results
are promising from the point of view of the versatility of
this material for optoelectronic applications
Authors’ website:.
J.F.S.R.,www.uv.es/lowdim
J.P.M.P., http://www.uv.es/umdo
2
Electronic structure, optical properties, and lattice dynamics in atomically thin Indium Selenide flakes
Juan F. Sánchez-Royo1 (), Guillermo Muñoz-Matutano1,†, Mauro Brotons-Gisbert1, Juan P. Martínez-Pastor1 (), Alfredo Segura1,2, Andrés Cantarero1, Rafael Mata1, Josep Canet-Ferrer1, Gerard Tobias3, Enric Canadell3, Jose Marqués-Hueso4, and Brian D. Gerardot4
1 ICMUV, Instituto de Ciencia de Materiales, Universidad de Valencia, P.O. Box 22085, 46071 Valencia, Spain. 2 MALTA-Consolider Team, Institut de Ciència dels Materials-Dpto. de Física Aplicada, Universitat de València, E-46100 Burjassot
(València), Spain. 3 Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus de la UAB, 08193 Bellaterra, Barcelona, Spain. 4 Institute of Photonics and Quantum Science, SUPA, Heriot-Watt University, Edinburgh EH14 4AS, UK.
† Present address: Optics and Quantum Communications group, ITEAM, UPV, Valencia, Spain.
Received: day month year / Revised: day month year / Accepted: day month year (automatically inserted by the publisher)
© Tsinghua University Press and Springer-Verlag Berlin Heidelberg 2011
ABSTRACT The progressive stacking of chalcogenide single layers gives rise to two-dimensional semiconducting
materials with tunable properties that can be exploited for new field-effect transistors and photonic devices.
Yet the properties of some members of the chalcogenide family remain unexplored. Indium selenide (InSe) is
attractive for applications due to its direct bandgap in the near infrared, controllable p- and n-type doping
and high chemical stability. Here, we reveal the lattice dynamics, optical and electronic properties of
atomically thin InSe flakes prepared by micro-mechanical cleavage. Raman active modes stiffen or soften in
the flakes depending on which electronic bonds are excited. A progressive blue-shift of the
photoluminescence peaks is observed for decreasing flake thickness (as large as 0.2 eV for three single-layers).
First principles calculations predict an even stronger increase of the bandgap, 0.40 eV, for three single layers,
and as much as 1.1 eV for a single layer. These results are promising from the point of view of the versatility
of this material for optoelectronic applications at the nanometer scale and compatible with Si and III-V
technologies.
KEYWORDS
Indium selenide, two-dimensional flakes, micro-Raman spectroscopy, Micro-Photoluminescence, Electronic
structure.
Nano Res DOI (automatically inserted by the publisher)
Research Article
————————————
Address correspondence to Juan F. Sánchez-Royo, [email protected]; Juan P. Martínez-Pastor, [email protected]
3
1 Introduction
The difficulties found to create semiconducting
graphene [1,2], while still maintaining the
extraordinary properties of semimetallic graphene
[3,4], constitute a major obstacle for the realization of
field effect transistors and functional junctions for
optoelectronics based on truly two-dimensional (2D)
materials. This has partly promoted increasing
research on layered dichalcogenide semiconductors
[5,6] that takes advantage of the weak interlayer
interactions to isolate, by graphene-like exfoliation
[7-10] or other techniques [11-13], either single or few
semiconducting layers. Particular optical properties
of these layers, different from the bulk, have already
been exploited in the first optoelectronic devices
already prepared, given their promising applications
in valleytronics and spintronics [14-16]. Bulk Mo and
W dichalcogenides (MoS2, MoSe2, WS2, and WSe2)
are indirect semiconductors, but single layers have a
direct character and become strongly
photoluminescent [17-19]. They have enabled the
realization of high-performance electric field effect
transistors [8-11] and photodetectors [20,21].
Bearing in mind these perspectives, it is not
surprising that research on 2D semiconductors
started to spread over other chalcogenides. In this
sense, much less is known about properties and
potential applications that layered III-VI
chalcogenides, such as GaSe and InSe, may have
when these semiconductors are reduced to the
nanoscale. Bulk GaSe is a relatively wide bandgap
semiconductor (~2.05 eV at room temperature [22])
that previously attracted some interest due to its
non-linear optical properties [23-25]. Bulk InSe is a
direct bandgap semiconductor (~1.25 eV at room
temperature [26]) that is appealing due to its optical
properties [23,24], the changes of its electronic
structure under pressure [27,28], the possibility of
preparing high-quality thin films [29], the
application of InSe for Li-batteries [30], and
photovoltaic devices that reached an efficiency of
11% [31]. Among these III-VI semiconductors, thin
GaSe platelets have been the first to show a
size-related blue-shift of the absorption edge that,
starting at ~2.0 eV in the bulk, enters into the UV for
a single layer [32]. The first potential applications of
few layer GaSe flakes seem to follow the path set out
by other layered chalcogenides. Mechanically
exfoliated GaSe [33] and InSe [34] flakes have been
successfully used as photodetectors with
outstanding responsivity and quantum efficiency.
Also, high-performance field effect transistors based
on single layers of GaSe have been fabricated [35].
Very recently, a 0.2 eV blue-shift in the absorption
edge due to confinement effects was observed in
mechanically exfoliated InSe flakes [36]. This
blue-shift with decreasing flake thickness was
accompanied with a sharply decreasing
photoluminescence (PL), which was attributed to a
direct-to-indirect bandgap crossover [36].
While initial results on InSe flakes are quite
promising, a further extension of experimental
results to even thinner films and theoretical support
are required to understand in detail how
confinement effects alter the electronic, vibrational,
and optical properties of the flakes. In the present
paper we corroborate the findings reported in Ref.
[36] regarding the significant blue-shift of the
absorption band edge by using confocal micro-PL
(-PL) measurements at 4 K on atomically thin InSe
flakes. We also observe a strong decrease of the PL
signal. These results are supported by our
first-principles calculations of the electronic structure
on strictly 2D InSe and on InSe flakes until ten single
layers thick. The 2D material, as the bulk, shows a
direct character of the bandgap, however, thin flakes
with thickness ranging between two and ten layers
thick show an indirect character with a valence band
maximum located only 15 meV above valence band
states at the -point, where the minimum of the
conduction band appears. The rather flat
valence-band dispersion around the -point in these
indirect flakes makes that electrons can easily be
transferred between energetically near states with a
small amount of thermal energy. These results
indicate that the observed strong decrease of the PL
signal is mainly related to the enhancement of
non-radiative recombination processes in the thin
flakes rather than to the evoked direct-to-indirect
4
bandgap crossover [36]. Our first-principles
calculations also demonstrate that states defining the
bandgap in the flakes and bulk have the same orbital
origin, predicting a bandgap increase of 1.14 eV and
0.4 eV for single and triple layer (TL) flakes,
respectively. Furthermore, we have carried out
Raman micro-spectroscopy at room temperature on
InSe flakes as thin as one TL. In contrast to previous
results reported [36], we achieve to observe a shift of
the InSe Raman modes in ultrathin flakes, whose
sign depends of the excited vibration mode. These
results evidence an electronic charge rearrangement
occurring in these flakes that, according to
first-principles calculations, strengthen Se-In bonds
in the detriment of In-In bonds.
2 Results and discussion
2.1 Structure and morphology of the InSe flakes
Figure 1 shows the atomic structure of the
semiconductor and an example of flakes obtained
by mechanical exfoliation. A single layer of InSe
(Fig. 1(a)) consists of four monoatomic sheets of
hexagonally arranged atoms tetrahedrally linked in
the sequence Se-In-In-Se via covalent bonds. In bulk,
these individual InSe layers are held together by
van der Waals forces, following a stacking sequence
in which the projection of the Se atoms of one layer
lies at the center of the triangle defined by the Se
atoms of the next layer. Among the different
polytypes permitted by this stacking sequence, InSe
crystallizes, at ambient conditions, in a
rhombohedral layered phase (InSe-I) -known as
-polytype- which belongs to the space group R3m.
The hexagonal non-primitive cell of the -polytype
is formed by a TL stack whose lattice parameters
are c=24.946 Å and a=b=4.002 Å [37]. The layered
structure of InSe favors the preparation of
atomically thin flakes by mechanical exfoliation as
shown in Fig. 1(b). As in graphene flakes prepared
on SiO2/Si substrates [3], the variation of the optical
contrast in a microscope image under visible
illumination can be related to nanometric changes
in the thickness of the InSe flake (see Figs. S1-S3 in
the Electronic Supplementary Material (ESM)),
which indicates that the flake of Fig. 1(b) is
exhibiting a multi-terrace thickness with the
thinnest parts of the flake being the most
translucent ones. Atomic force microscopy (AFM)
results (Figure 1(c)) show that the flake is formed by
wide terraces of uniform thickness ranging between
2.5 and 12 nm (1 and 5 TL, respectively), which can
be univocally identified with the color-contrast
areas observed in the optical image.
2.2 Vibrational properties
As -InSe belongs to the R3m space group with four
atoms per primitive unit cell, there are twelve normal
modes of vibration. Group-theoretical considerations
[38] lead to decomposition at the point as 4A1+4E.
Apart from the acoustic ones (A1+E) all modes are
Raman and infrared active, but with very different
Figure 1 (a) Crystal structure of -InSe, in which the axes of
the hexagonal unit cell are plotted. The corresponding
hexagonal Brillouin zone has also been included. (b) Optical
image of a multi-terraced InSe flake exfoliated onto a SiO2/Si
substrate. Each color areas of the image can be associated to a
single terrace. (c) AFM image of the InSe flake. The thickness
profile measured along the marked ABCD path is shown at the
bottom of the image. For the sake of clarity we have indicated,
in this plot, the height expected for the stacking of single -and
triple- layers of InSe by horizontal dashed lines.
5
relative intensities. Two of them (A1+E) are strongly
polar and, consequently, more active in infrared,
developing a TO-LO splitting. Reported bulk-InSe
Raman spectra exhibit modes whose frequency,
symmetry, and vibration schemes are shown in Fig.
2(a) [39]. All these vibrational modes can be
recognized in the measured Raman of Fig. 2(b),
which were registered in a thick flake and in
different points of the multi-terrace flake (labeled
spots in the image inset of Fig. 2(b)). Raman
scattering intensity can be enhanced by using the
excitation laser photon energy resonant with the E’1
direct allowed transition at 2.4 eV in bulk InSe
[25,39,40]. Consequently, Raman peaks appear over a
PL tail associated to this optical transition, whose
intensity progressively decreases as the thickness of
the probed terrace becomes thinner. This is an
important observation since it is a clear indication of
the blue-shift of direct band-to-band optical
transitions. Furthermore, variations in the
asymmetry degree of the 175 cm-1 E mode (panel I in
Fig. 2(b)) and the strong decrease in intensity of the
second order Raman peaks (panel II in Fig. 2(b)),
from bulk-like to thin flakes, are also indicative of
the loss of resonance in the thinnest parts of the flake
[39].
In addition to the electronic blue-shift inferred from
E’1-resonant Raman spectra, the thickness reduction
of a flake also affects its lattice dynamics, as revealed
by the shift of the different phonon frequencies
observed with respect to the bulk (Fig. 2(c)). As the
flake thickness reduces, the frequency of non-polar
A1 modes slightly decreases while that of non-polar
E and polar A1(LO) modes increases significantly.
This shift of the phonon modes as flakes thickness
decreases becomes more evident in very thin flakes
(<10 nm), which would explain why this effect has
not been detected in a previous work reporting
Raman spectroscopy results on InSe thin flakes [34].
In order to understand the results showed in Fig. 2(c),
we have assessed the evolution of the different bond
strengths with the number of layers by calculating
the so-called Mulliken [41,42] bond overlap
populations for bulk -InSe and its single layer.
Intralayer In-Se and In-In bonds are strong covalent
bonds, while interlayer bonding is due to weaker van
der Waals interactions. For both InSe bulk and single
layer, we have estimated the bonding strength in two
spatial regions in which the interlayer electronic
density reaches its maximum: the line between
closest Se atoms from adjacent layers and the line
from an In atom in one layer and the nearest Se atom
in the adjacent layer (Table I). The In-Se bonds are
stronger in the single layer than in bulk InSe,
whereas the opposite is found for In-In bonds. Of
course, the interlayer interactions weaken and finally
disappear when going from the bulk to the single
layer. The relative variations of the In-Se and In-In
bond overlap populations due to the presence of
other layers is around 3% of the strength of a full
covalent bond. Although apparently small, these
overlap population changes would give rise to a
relative variation around 2% for the natural
frequency of a single oscillator, which would already
account for the magnitude of the observed relative
shift of the Raman peaks. In any case, it must be
outlined that our calculations have been performed
using exactly the same atomic geometries in all cases,
so the estimated changes of the bond overlap
population only contain purely electronic effects that
would be subsequently magnified upon structural
optimization, similar to what happens for other
layered materials [43].
From these estimates we can expect that those
phonon modes in which the In-Se bonds play an
important role (due to bond stretching in the
vibration scheme of Fig. 2(a)) should exhibit a
frequency increase in an isolated single layer: this is
Table 1 DFT bond overlap populations calculated for bulk and
one single layer of InSe.
Bond Intralayer Interlayer
In-Se In-In In-Se Se-Se
Bulk 0.313 0.315 0.019 0.004
Single layer 0.324 0.303 0 0
6
indeed the case of the non-polar E (175 cm-1) and
polar A1(LO) phonons. In low frequency phonons E
(40 cm-1) and A1 (114 cm-1), the In-Se bond does not
play a significant role (covalently bonded In and Se
vibrate in phase) so that restoring forces are mainly
guaranteed by interlayer interactions and by either
In-In bond bending (E) or stretching (A1). Then, the
disappearance of interlayer interactions and the
weakening of the In-In bond are consistent with the
frequency decrease of the low frequency phonons
observed when going from bulk to single layer. In
the case of the non-polar A1 mode at 225 cm-1 all
bonds provide restorative stretching force and its
frequency decrease can be explained by the fact that
the disappearance of interlayer forces and the (larger)
weakening of In-In bonds overcompensate the
strengthening of In-Se bonds.
2.3 Optical properties
Raman spectroscopy results revealed a blue-shift of
the direct transition E’1 connecting deep px-py valence
band states with conduction band states [25,38,39].
As these valence band states are rather localized in
the layer plane, the observed blue-shift should be a
consequence of the upshift of the conduction band as
the flake thickness decreases. A deeper analysis of
the confinement effects on the electronic properties
of InSe flakes can be performed by -PL techniques,
as reported for other layered materials
[5,13,17,18,44,45]. In bulk InSe, the low-temperature
PL signal is dominated by exciton and
impurity-related radiative recombination [46,47]. In
InSe flakes, hD0 recombination channels seem to
dominate over bound exciton recombination, as
Figure 2 (a) Vibrational structure of the detected first-order Raman active modes. (b) Left panel: Micro-Raman spectra measured under
bulk resonant conditions in the different terraces of the InSe flake labeled from a to g in the inserted image. The thickness of the probed
flake terrace is indicated on each Raman spectrum. The identified vibration modes have been labeled. Vertical dashed bars correspond
to the position that these vibration modes have in the bulk. We have included the spectrum measured in a bulk sample. We have
checked that the position of the SiO2-related Raman peak appears at 518 cm-1 in all acquired spectra. Panels I and II are zooms of the
corresponding marked-color areas in the main figure. For the sake of comparison, the photoluminescence tail has been subtracted in the
spectra shown in these two color-panels. The Raman peaks were fitted by a single Lorentzian curve (two in the case of the E mode at
175 cm-1 to account for its asymmetric line-shape). (c) Thickness-flake dependence of the relative frequency shift of the Raman modes
(with respect to bulk frequency) as obtained by Lorentzian fitting of the measured Raman peaks.
7
suggested by the comparison between the -PL
spectra measured at the thickest terrace (12 nm thick
terrace, marked as a-point in the image inset of Fig.
3(a)) and the bulk sample (Fig. S4 in the ESM). This
assignment is also supported by the analysis of the
-PL transient measured in thick InSe flakes (Figs.
S5-S6 in the ESM). Figure 3(a) shows the -PL spectra
measured at T = 4 K in the different terraces of the
flake corresponding to the points labeled from a to f
on the image at the inset. As aforementioned, hD0
transition dominates the -PL band recorded at
a-point and hence matches the low-energy tail of the
bulk PL [47]. As the collection spot moves close to
the next terrace, b-point, new components appear
towards the high-energy side of the -PL spectrum,
with the highest energy one (at 1.35 eV) clearly
separated from the main feature observed at lower
energies and located above the free-exciton
recombination line in bulk (at 1.337 eV) [47]. As the
collection spot travels to next thinner sections of the
flake (points c to f), the whole -PL spectrum shifts
undoubtedly to the blue (Figure 3(a)) with a
Figure 3 (a) Low temperature -PL spectra measured in the different terraces of the InSe flake labeled from a to f in the image at the
top. The thickness of the probed flake terrace is indicated on each PL spectrum. Solid red lines are the result of the fitting to the
experimental spectra by using several Gaussian curves of the same width. Deconvolution of the fitting curves into their single Gaussian
components is shown under each PL spectrum. A Color-code has been used to identify similar Gaussian components that result from
the deconvolution of the different PL spectra acquired. Note that the spectra measured in the 5 and 2 nm-thick terraces have been
magnified. (b) The left panel shows the maximum energy of the six different Gaussian components (labeled as from P1 to P6 as
increasing in energy) which have been found to conform to the measured PL spectra. The right panel shows the intensity of these fitted
Gaussian components. These two panels are plotted as a function of the spot position. The color and symbol codes have been
maintained in all plots included in (a) and (b). (c) Black circles: Maximum energy of Gaussian components as a function of the inverse
square of the assigned number of InSe layers. Red dashed line: Predicted bandgap of an InSe flake as a function of the number of
layers, as given by the effective mass model. White open circles: bandgap of InSe flakes as a function of the number of layers, as
predicted by first-principles band structure calculations. Blue triangles: experimental data extracted from Ref. [36].
8
lineshape that depends on the terrace scanned. At
the same time, the -PL spectra measured at these
points of the flake experience a strong decrease in
intensity, whose origin is not completely clear from
an experimental point of view and some hypothesis
will be proposed afterwards.
These results indicate the -PL signal shifts to the
blue, as the terrace thickness decreases, due to the
progressive development of higher energy
components that become dominant at the
consecutive thin parts of the flake. In order to get
insight into the analysis of these -PL spectra, a
multi-Gaussian deconvolution has been performed
by assuming all the discrete -PL components have
the same line width. Proceeding in this way, a
minimum number of six components, labeled as
P1-P6 (Figures 3(a) and 3(b)), have been found to
reproduce all -PL spectra acquired. These -PL
components are thus associated to the decreasing
thickness of the terraces from a to f points as
originated from quantum confinement effects.
Notably, multi-component -PL spectra are detected
across the flake, even when the collection spot
probes a unique flake terrace, as shown and
indicated in Figs. 3(a) and 3(b) (left panel),
respectively. In the -PL measurements the excitation
area is much larger than the collection spot (see
Experimental methods subsection), suggesting that
the multi-component -PL spectra are due to carriers
photo-excited in a particular point of the flake
drifting and recombining at the edges of the terrace.
We also observe an extremely sharp decrease of the
integrated intensity of P1-P6 components (right
panel in Figure 3(b)), which contrasts to the expected
linear trend for such thin layers. This would suggest
an enhancement of the influence of non-radiative
surface carrier recombination processes as the
thickness decreases. However, a larger density of
traps or surface defects in the thinner terraces is
rather unexpected on the basis that the surfaces of
the different terraces of the flake have suffered the
same history. -PL intensity maps measured at room
temperature (Fig. 1 in Ref. [36]) reflect an
inhomogeneous intensity distribution, accumulated
towards the edges of the terraces, which would be
expected to become more pronounced at low
temperatures (due to the higher mobility of carriers)
and in thinner terraces (due to the unavoidable
existence of recombination traps at the surface of the
SiO2 substrate underneath). Therefore, the behavior
of the -PL intensity observed in Fig. 3(b) points out,
again, that radiative recombination tends to occur at
the edge of the terraces (see also Fig. S7 in the ESM)).
All these remarks can be understood taking into
account that, really, multi-terrace InSe flakes are
expected to behave as a planar array of
heterojunctions formed by semiconductors with
different bandgap values and diffusion length of
carriers as large as that of the bulk [48]. In such
devices, photo-excited carriers in thinner terraces
(with larger band gap values due to strong
confinement effects) diffuse and recombine at the
terrace edge, whereas carriers excited in the thicker
terraces would eventually recombine further inside
neighboring terraces of lesser thickness, provided
that confinement effects are soft enough in these
thick terraces to allow lateral coupling of electronic
states at their edges. This picture would also explain
the lack of -PL signal observed in isolated very thin
flakes (2-4 nm thick) of uniform thickness (Fig. S7 in
the ESM): In these very thin flakes, mechanisms
evoked above to promote diffusion of photo-excited
carriers towards the flake edge are no longer acting
(no planar array of heterojunctions there exists).
Under these assumptions isotropic diffusion would
favor surface recombination, which is dominated by
non-radiative processes (as recombination at surface
defects and/or traps at the SiO2 substrate).
2.4 Electronic properties
The present results offer a remarkable opportunity to
study how the progressive enhancement of quantum
size confinement can strongly tune the electronic
properties of these atomically thick semiconductor
flakes. The hypothesis made above to assign each
-PL component to a flake thickness value would not
be straightforward, but the limited number of
thickness steps in the flake (Figure 1(c)) allows for a
reliable assignment. On the assumption that the
9
highest energy component corresponds to the
thinnest part of the flake (1 TL), the assignment
sequence of the P1-P6 components will be 12 (4 TL),
10, 8, 6 (2 TL), 4 and 3 (1 TL) single layers,
respectively. This assignment would tell us that the
recombination energy of impurity-related states
increases by 0.2 eV when the flake becomes as thin as
1 TL (Figure 3(c)).
An effective mass model assuming the flake as a
quantum well with infinite potential barriers of
electronic confinement (Figure 3(c)) approximately
reproduces the already reported PL results at room
temperature in flakes thicker than few TL [36] and
-PL peak energies obtained at the thicker flakes
studied in this work. However, the 1/N2 dependence
(N is the number of single layers) expected from this
simple model fails to explain results in the thinnest
flakes. On the one hand, a more refined study of the
confinement effect on the electronic properties of the
flakes is necessary in order to quantitatively
understand experimental results beyond the effective
mass model. On the other hand, the band structure
calculation based on first-principles as a function of
the number of InSe single layers is the basis to
understand future works on optical properties of
flakes prepared with this material.
First-principles band structures calculated for bulk
-InSe and flakes containing eight, four, and one InSe
single layers are shown in Fig. 4 (calculations were
carried out for one to ten single layers). For ease of
comparison, the band structure of bulk -InSe has
been represented in the Brillouin zone of the
hexagonal unit cell (Figure 1(a)) instead of the
rhombohedral one. Note that in the band structures
of Figure 4(b)-(d) the A direction has no meaning,
and that band diagrams in the -M-K and A-L-H
Figure 4 DFT band structure calculated for (a) bulk -InSe and for (b)-(d) 8, 4, and 1 single InSe layers, respectively, through high
symmetry directions of the hexagonal Brillouin zone (Figure 1(a)). In all cases, the energy zero has been placed at the top of the highest
occupied band. The bandgap has been identified in the plots by a green shaded area.
10
planes are identical since the strongly dispersing
three-dimensional bands of the bulk become a series
of N 2D bands (N=8,4,1 in Figure 4(b)-(d),
respectively). The anisotropic character of bulk
-InSe band structure is mainly reflected in the
strong difference between bands with Se px-py orbital
character, with no dispersion along the A direction,
and those with mainly Se pz, In pz and In s character,
that exhibit strong dispersion along the A direction.
The latter bands form the uppermost valence band,
with nonbonding Se pz character, and the lowermost
conduction band, with mainly antibonding In s
character. The main orbital character of these bulk
bands does not change in the flakes. In fact, note that
the upper filled and lower empty bands of the bulk,
i.e. those in the A-L-H plane of the Brillouin zone,
have the same shape as the upper valence and lower
conduction bands of the flakes with the only
difference of the progressive reduction of the flat
region around in the upper filled band. This is in
contrast to observations in flakes based on layered
transition-metal dichalcogenides, in which the
orbital origin of states defining the bandgap differs
from the bulk [17,18,44,45,49]. It is also relevant to
notice that single layer InSe is a direct semiconductor
(Figure 4(d)). However, in few layer InSe (Figure
4(a)-(c)), the valence band maximum occurs slightly
away from the point, at 1/10 of the K and M
directions, and around 15 meV above the point
energy. This change is related to the ring-shaped
valence band maximum emerging in bulk InSe under
pressure, which has been shown to contribute to the
quenching of the exciton absorption peak [27] but
has little effect on the intrinsic photoluminescence
[50]. It seems unlikely that this change in the valence
band maximum is responsible for the PL quenching
observed in Ref. [36], because the energy difference
with respect to is one half of the thermal energy at
room temperature.
Since the most visible effect of confinement on the
bulk electronic band structure occurs for the valence
and conduction bands, density functional theory
(DFT) calculations predict a bandgap increase in the
flakes which can be as high as 0.40 eV for one TL and
1.14 eV for a single layer, with respect to the bulk.
Figure 3(c) includes the results of these calculations,
using as a reference the value of the bulk bandgap at
1.35 eV at low temperatures [26] (note that DFT
calculations usually underestimate absolute bandgap
values). The hD0 recombination and the calculated
bandgap energies seem to follow a quite similar
trend with the thickness (fast increase for 1/N2<0.04
and smooth increase above this value), although a
lower blue-shift of the hD0 recombination energy
seems to occur (0.2 eV for one TL) attributable to the
shorter spatial extension of the deep donor states
involved in these optical transitions. For relatively
thick flakes (12-8 nm thick), confinement effects
mostly act on extended states defining the InSe
bandgap. In these relatively thick flakes,
impurity-related states can still be considered as bulk
ones and seem to evolve with the flake thickness as
electronic states determining the bandgap evolution.
However, for thinner flakes, confinement effects
already acting on donor-related states progressively
increase the donor activation energy similarly to
what occurs with bulk hydrogenic (shallow) excitons
[51]: When bulk excitons become localized, they go
deeper into the bandgap due to the fact that,
although they turn into 2D, Coulomb interactions
always remain three-dimensional [49,52,53].
3 Conclusions
In summary, we have demonstrated that the gradual
enhancement of quantum size confinement has
enabled tuning the dimensionality of the lattice
dynamics and electronic properties of atomically thin
InSe flakes. In this way, electronic charge rearranges
and Raman modes involving quasi-in-plane bonds
stiffen whereas those involving bonds perpendicular
to the layer soften. Of course, the most evident effect
of lateral confinement appears on the electronic
properties of extended states. DFT calculations
predict a huge increase of the electronic bandgap by
more than 1 eV for a single InSe layer (2D case). The
relative increase of the bandgap calculated by DFT in
relatively thick flakes nicely reproduces the increase
in energy of the -PL band that was assigned to hD0
transitions associated to donors in interlayer sites.
Quantum size confinement effects do not alter the
11
orbital nature of states determining the bandgap,
thus InSe arises as a relative simple model system to
study the dynamics of confined carriers and
impurity-related states. These results also suggest
that InSe has potential applications in future planar
electronic and optoelectronic nanodevices
compatible with Si and III-V technologies, in
combination with other planar materials to create
top-down heterostructures, and eventually using
graphene as a transparent electrode [54,55]. For these
devices, it is worth noting that InSe can be
intentionally n-type (introducing Si and Sn
impurities) [46,56] and p-type doped (introducing
transition metals, as Zn and Pt) [57,58]. Also, the
doping versatility of this material suggests that 2D
systems with properties difficult to combine in the
bulk can be achieved, as the recently reported InSe
nanocrystals exhibiting room-temperature
ferromagnetism when doped with Fe and Co [59].
Finally, it should be remarked that, given the
intrinsic exhaustion of covalent bonds in the atomic
single layers of chalcogenide families and their
natural stacking by Van der Waals forces, one can
infer that heterostructures formed by different
2D/quasi-2D materials are possible and it will
constitute a predictable field of research in next
future from both basic and applications points of
view.
4 Methods
4.1 Experimental methods
InSe monocrystals here used to prepare the flakes
were cleaved perpendicular to the (001) direction
from an ingot grown by the Bridgman method from
a nonstoichiometric In1.05Se0.95 melt in which Tin, in a
content 0.01%, was introduced previously to growth
to act as n-dopant [60]. From these ingots, thin
n-doped InSe samples were cleaved and used to
prepare atomically thin InSe flakes on SiO2/Si
substrates by means of a mechanical exfoliation
technique similar to that employed for graphene [3].
The thickness of the SiO2 layer was 300 nm, which
potentially makes few-layer InSe flakes optically
detectable (see Figs. S1-S3 in the ESM). The thickness
of the InSe flakes was measured by AFM. AFM
imaging was performed by using a Nanotec
multimode microscope operating in non-contact
mode.
For -PL measurements a CW 405 nm laser was used
as excitation source (Power ~23 W). The laser beam
is coupled to a monomode optical fiber and carried
to the excitation arm of a diffraction limited confocal
microscope, working at liquid helium temperature.
The collection spot size on the sample was less than 1
m. The -PL signal is coupled to the collection
optical fiber, and was attached to a double 0.3 m
monochromator. An ANDOR technology charge
coupled device (CCD) was attached to the
monochromator front output to record the -PL
spectra.
Vibrational properties of the InSe flakes were studied
by Raman spectroscopy. Raman spectra were
collected at room temperature with a JY-Horiba
T64000 Raman microprobe spectrometer in
backscattering configuration (laser spot diameter ~1
m, objective numerical aperture = 0.90). As
excitation source, the 514.53 nm line of an Ar+ laser
was used and its plasma peaks served as fixed
frequency references. The detector is a liquid
N2-cooled CCD. The power at the sample was
limited to 100 W, in order to limit heating effects
and to avoid any sample damage.
4.2 Electronic structure calculations
Band structure calculations were carried out using a
numerical atomic orbitals DFT approach [61,62],
which has been developed and designed for efficient
calculations in large systems and implemented in the
SIESTA code [63-65]. We have used the local density
approximation to DFT and, in particular, the
functional of Perdew and Zunger [66]. Only the
valence electrons are considered in the calculation,
with the core being replaced by norm-conserving
scalar relativistic pseudopotentials [67] factorized in
the Kleinman-Bylander form [68]. Nonlinear partial
core corrections to describe the exchange and
correlations in the core region were included [69].
12
We have used a split-valence double- basis set
including polarization orbitals for all atoms as
obtained with an energy shift of 250 meV [70]. The
energy cutoff of the real space integration mesh was
300 Ry and the Brillouin zone was sampled using
grids of (10x10x1) and (10x10x10) k-points [71] for
the different slabs and the bulk, respectively. The
calculations for the different slabs (from 1 to 10
layers) were carried out using the same geometrical
details as for the bulk [37] since Rybkovskiy et al. [32]
have recently shown that the geometrical variations
between the bulk and different slabs of GaSe are not
substantial.
Acknowledgements
This work was supported by the Spanish
Government (Grants FIS2012-37549-C05-05,
MAT2011-24757, CSD2007-00041,
MAT2012-38664-C02-02, CSD2007-00045, and
TEC2011-29120-C05-01/-05), the Comunidad
Valenciana Government (Grant
PROMETEO/2009/074) and the Royal Society, the
European Research Council, and the Engineering
and Physical Sciences Research Council (UK).
G.M.M. acknowledges his national fellowship no.
JCI-2011-10686 under Juan de la Cierva program.
Electronic Supplementary Material:
Supplementary material information on InSe thin
flakes identification processes and on low
temperature photoluminescence spectra acquired in
bulk InSe samples and single-terraces flakes is
available in the online version of this article at
http://dx.doi.org/10.1007/s12274-***-****-*.
References
[1] Lu, G.; Yu, K.; Wen, Z.; Chen, J. Semiconducting graphene:
converting graphene from semimetal to semiconductor.
Nanoscale 2013, 5, 1353-1368.
[2] Wang, F.; Liu, G.; Rothwell, S.; Nevius, M.; Tejeda, A.;
Taleb-Ibrahimi, A.; Feldman, L. C.; Cohen, P. I.; Conrad, E.
H. Wide-gap semiconducting graphene from
Nitrogen-seeded SiC. Nano Lett. 2013, 13, 4827-4832.
[3] Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.;
Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A.
Electric field effect in atomically thin Carbon films. Science
2004, 306, 666-669.
[4] Wang, Y.; Brar, V. W.; Shytov, A. V.; Wu, Q.; Regan, W.;
Tsai, H.-Z.; Zettl, A.; Levitov, L. S.; Crommie, M. F.
Mapping Dirac quasiparticles near a single Coulomb
impurity on graphene. Nature Phys. 2012, 8, 653-657.
[5] Wang, Q. H.; Kalantar-Zadeh, K.; Kis, A.; Coleman, J. N.;
Strano, M. S. Electronics and optoelectronics of
two-dimensional transition metal dichalcogenides. Nature
Nanotech. 2012, 7, 699-712.
[6] Geim, A. K.; Grigorieva, I. V. Van der Waals
heterostructures. Nature 2013, 499, 419-425.
[7] Ayari, A.; Cobas, E.;, Ogundadegbe, O.; Fuhrer, M. S.
Realization and electrical characterization of ultrathin
crystals of layered transition-metal dichalcogenides. J. Appl.
Phys. 2007, 101, 014507.
[8] Zhang, Y.; Ye, J.; Matsuhashi, Y.; Iwasa, Y. Ambipolar
MoS2 thin flake transistors. Nano Lett. 2012, 12, 1136-1140.
[9] Fang, H.; Chuang, S.; Chang, T. C.; Takei, K.; Takahashi, T.;
Javey, A. High-performance single layered WSe2 p-FETs
with chemically doped contacts. Nano Lett. 2012, 12,
3788-3792.
[10] Braga D.; Gutiérrez Lezama, I.; Berger, H.; Morpurgo, A. F.
Quantitative determination of the band gap of WS2 with
ambipolar ionic liquid-gated transistors. Nano Lett. 2012, 12,
5218-5223.
[11] Hwang, W. S.; Remskar, M.; Yan, R.; Protasenko, V.; Tahy,
K.; Chae, S. D.; Zhao, P.; Konar, A.; Xing, H.; Seabaugh, A.;
Jena, D. Transistors with chemically synthesized layered
semiconductor WS2 exhibiting 105 room temperature
modulation and ambipolar behavior. Appl. Phys. Lett. 2012,
101, 013107.
[12] Liu, K. K.; Zhang, W.; Lee, Y.-H.; Lin, Y.-C.; Chang, M.-T.;
Su, C.-Y.; Chang, C.-S.; Li, H.; Shi, Y.; Zhang, H.; Lai,
C.-S.; Li, L.-J. Growth of large-area and highly crystalline
MoS2 thin layers on insulating substrates. Nano Lett. 2012,
12, 1538-1544.
13
[13] Elias, A. L.; Perea-López, N.; Castro-Beltrán, A.;
Berkdemir, A.; Lv, R.; Feng, S.; Long, A. D.; Hayashi, T.;
Kim, Y. A.; Endo, M.; Gutiérrez, H. R.; Pradhan, N. R.;
Balicas, L.; Mallouk, T. E.; López-Urías, F.; Terrones, H.;
Terrones, M. Controlled synthesis and transfer of large-area
WS2 sheets: from single layer to few layers. ACS Nano 2013,
7, 5235-5242.
[14] Mak; K. F.; He, K.; Shan, J.; Heinz, T. F. Control of valley
polarization in monolayer MoS2 by optical helicity. Nature
Nanotech. 2012, 7, 494-498.
[15] Zeng, H.; Dai, J.; Yao, W.; Xiao, D.; Cui, X. Valley
polarization in MoS2 monolayers by optical pumping. Nature
Nanotech. 2012, 7, 490-493.
[16] Cao, T.; Wang, G.; Han, W.; Ye, H.; Zhu, C.; Shi, J.; Niu,
Q.; Tan, P.; Wang, E.; Liu, B.; Feng, J. Valley-selective
circular dichroism of monolayer molybdenum disulphide.
Nature Commun. 2012, 3, 887.
[17] Mak, K. F.; Lee, C.; Hone, J.; Shan, J.; Heinz, T. F.
Atomically thin MoS2: a new direct-gap semiconductor. Phys.
Rev. Lett. 2010, 105, 136805.
[18] Tongay, S.; Zhou, J.; Ataca, C.; Lo, K.; Matthews, T. S.; Li,
J.; Grossman, J. C.; Wu, J. Thermally driven crossover from
indirect toward direct bandgap in 2D semiconductors: MoSe2
versus MoS2. Nano Lett. 2012, 12, 5576-5580.
[19] Zhao, W.; Ghorannevis, Z.; Chu, L.; Toh, M.; Kloc, C.; Tan,
P.-H.; Eda, G. Evolution of electronic structure in atomically
thin sheets of WS2 and WSe2. ACS Nano 2013, 7, 791-797.
[20] Lopez-Sanchez, O.; Lembke, D.; Kayci, M.; Radenovic, A.;
Kis, A. Ultrasensitive photodetectors based on monolayer
MoS2. Nature Nanotech. 2013, 8, 497-501.
[21] Baugher, B. W. H.; Churchill, H. O. H.; Yang, Y.;
Jarillo-Herrero, P. Optoelectronic devices based on
electrically tunable p-n diodes in a monolayer dichalcogenide.
Nature Nanotech. 2014, 9, 262-267.
[22] Gauthier, M.; Polian, A.; Besson, J. M.; Chevy, A. A.
Optical properties of Gallium Selenide under high pressure.
Phys. Rev. B 1989, 40, 3837-3854.
[23] Bringuier, E.; Bourdon, A.; Piccioli, N.; Chevy, A. Optical
second-harmonic generation in lossy media: Application to
GaSe and InSe. Phys. Rev. B 1994, 49, 16971-16982.
[24] Segura, A.; Bouvier, J.; Andrés, M. V.; Manjón, F. J.;
Muñoz, V. Strong optical nonlinearities in gallium and
indium selenides related to inter-valence-band transitions
induced by light pulses. Phys. Rev. B 1997, 56, 4075-4084.
[25] Ferrer-Roca, Ch.; Bouvier, J.; Segura, A.;, Andrés, M. V.;
Muñoz, V. Light-induced transmission nonlinearities in
gallium selenide. J. Appl. Phys. 1999, 85, 3780-3785.
[26] Camassel, J.; Merle, P.; Mathieu, H.; Chevy, A. Excitonic
absorption edge of indium selenide. Phys. Rev. B 1978, 17,
4718-4725.
[27] Manjón, F.J.; Errandonea, D.; Segura, A.; Muñoz, V.;
Tobías, G.; Ordejón, P.; Canadell, E. Experimental and
theoretical study of band structure of InSe and In1-xGaxSe
(x<0.2) under high pressure. Phys. Rev. B 2001, 63, 125330.
[28] Errandonea, D.; Segura, A.; Manjón, F. J.; Chevy, A.;
Machado, E.; Tobias, G.; Ordejón, P.; Canadell, E. Crystal
symmetry and pressure effects on the valence band structure
of γ-InSe and ϵ-GaSe: Transport measurements and
electronic structure calculations. Phys. Rev. B 2005, 71,
125206.
[29] Sánchez-Royo, J. F.; Segura, A.; Lang, O.; Schaar, E.;
Pettenkofer, C.; Jaegermann, W.; Roa, L.; Chevy, A. Optical
and photovoltaic properties of indium selenide thin films
prepared by van der Waals epitaxy. J. Appl. Phys. 2001, 90,
2818-2823.
[30] Julien, C.; Balkanski, M. Thin-film growth and structure for
solid-state batteries. Appl. Surf. Sci. 1991, 48-49, 1-11.
[31] Martinez-Pastor, J.; Segura, A.; Valdes, J. L.; Chevy, A.
Electrical and photovoltaic properties of
indium-tin-oxide/p-InSe/Au solar cells. J. Appl. Phys. 1987,
62, 1477-1483.
[32] Rybkovskiy, D. V.; Arutyunyan, N. R.; Orekhov, A. S.;
Gromchenko, I. A.; Vorobiev, I. V.; Osadchy, A. V.; Salaev,
E. Yu.; Baykara, T. K.; Allakhverdiev, K. R.; Obraztsova, E.
D. Size-induced effects in gallium selenide electronic
structure: The influence of interlayer interactions. Phys. Rev.
B 2011, 84, 085314.
[33] Hu, P.A.; Wen, Z.; Wang, L.; Tan, P.; Xiao, K. Synthesis of
few-layer GaSe nanosheets for high performance
photodetectors. ACS Nano 2012, 6, 5988-5994.
[34] Lei, S.; Ge, L.; Najmaei, S.; George, A.; Kappera, R.; Lou,
J.; Chhowalla, M.; Yamaguchi, H.; Gupta, G.; Vajtai, R.;
Mohite, A. D.; Ajayan, P. M. Evolution of the electronic
14
band structure and efficient photo-detection in atomic layers
of InSe. ACS Nano 2014, 8, 1263-1272.
[35] Late, D. J.; Liu, B.; Luo, J.; Yan, A.; Ramakrishna Matte, H.
S. S.; Grayson, M.; Rao, C. N. R.; Dravid, V. P. GaS and
GaSe ultrathin layer transistors. Adv. Mater. 2012, 24,
3549-3554.
[36] Mudd, G. W.; Svatek, S. A.; Ren, T.; Patanè, A.;
Makarovsky, O.; Eaves, L.; Beton, P. H.; Kovalyuk, Z. D.;
Lashkarev, G. V.; Kudrynskyi, Z. R.; Dmitriev, A. I. Tuning
the bandgap of exfoliated InSe nanosheets by quantum
confinement. Adv. Mater. 2013, 25, 5714-5718.
[37] Rigoult, J.; Rimsky, A.; Kuhn, A. Refinement of the 3R
γ-Indium Monoselenide structure type. Acta Cryst. B 1980,
36, 916-918.
[38] Faradev, F. E.; Gasanly, N. M.; Mavrin, B. N.; Melnik, N. N.
Raman scattering in some III-VI layer single crystals. Phys.
Stat. Sol. (b) 1978, 85, 381-386.
[39] Kuroda N.; Nishina, Y. Resonant Raman scattering at higher
M0 exciton edge in layer compound InSe. Solid State
Commun. 1978, 28, 439-443.
[40] Millot, M.; Broto, J.-M.; George, S.; González, J.; Segura,
A. Electronic structure of indium selenide probed by
magnetoabsorption spectroscopy under high pressure. Phys.
Rev. B 2010, 81, 205211.
[41] Mulliken, R. S. Electronic population analysis on
LCAO-MO molecular wave functions. I. J. Chem. Phys.
1955, 23, 1833-1840.
[42] Mulliken, R. S. Electronic population analysis on
LCAO-MO molecular wave functions. II. J. Chem. Phys.
1955, 23, 1841-1846.
[43] Arenal, R.; Ferrari, A. C.; Reich, S.; Wirtz, L.; Mevellec,
J.-Y.; Lefrant, S.; Rubio, A.; Loiseau, A. Raman
Spectroscopy of Single-Wall Boron Nitride Nanotubes. Nano
Lett. 2006, 6, 1812-1816.
[44] Splendiani, A.; Sun, L.; Zhang, Y.; Li, T.; Kim, J.; Chim,
C.-Y.; Galli, G.; Wang, F. Emerging photoluminescence in
monolayer MoS2. Nano Lett. 2010, 10, 1271-1275.
[45] Zeng, H.; Liu, G.-B.; Dai, J.; Yan, Y.; Zhu, B.; He, R.; Xie,
L.; Xu, S.; Chen, X.; Yao, W.; Cui, X. Optical signature of
symmetry variations and spin-valley coupling in atomically
thin tungsten dichalcogenides. Sci. Rep. 2013, 3, 1608.
[46] Riera, J.; Segura, A.; Chevy, A. Photoluminescence in
silicon-doped n-indium selenide. Phys. Stat. solidi (a) 1994,
142, 265-274.
[47] Ferrer-Roca, Ch.; Segura, A.;, Andrés, M. V.; Pellicer, J.;
Muñoz, V. Investigation of nitrogen-related acceptor centers
in indium selenide by means of photoluminescence:
Determination of the hole effective mass. Phys. Rev. B 1997,
55, 6981-6987.
[48] Segura, A.; Guesdon, J. P.; Besson, J. M.; Chevy, A.
Photoconductivity and photovoltaic effect in indium selenide.
J. Appl. Phys. 1983, 54, 876-888.
[49] Ramasubramaniam, A. Large excitonic effects in
monolayers of molybdenum and tungsten dichalcogenides.
Phys. Rev. B 2012, 86, 115409.
[50] Manjón, F. J.; Segura, A.; Muñoz-Sanjosé, V.; Tobías, G.;
Ordejón, P.; Canadell, E. Band structure of indium selenide
investigated by intrinsic photoluminescence under high
pressure. Phys. Rev. B 2004, 70, 125201.
[51] Martinez-Pastor, J.; Segura, A.; Julien, C.; Chevy, A.
Shallow-donor impurities in indium selenide investigated by
means of far-infrared spectroscopy. Phys. Rev. B 1992, 46,
4607-4616.
[52] Davies, J. H. The Physics of Low-dimensional
Semiconductors: An Introduction. (Cambridge University
Press, New York, 1997).
[53] Dvorak, M.; Wei, S.-H.; Wu, Z. Origin of the variation of
exciton binding energy in semiconductors. Phys. Rev. Lett.
2013, 110, 016402.
[54] Youn, D.-H.; Yu, Y.-J.; Choi, H.; Kim, S.-H.; Choi, S.-Y.;
Choi, C.-G. Graphene transparent electrode for enhanced
optical power and thermal stability in GaN light-emitting
diodes. Nanotechnology 2013, 24, 075202.
[55] Lee, M. S.; Lee, K.; Kim, S.-Y.; Lee, H.; Park, J.; Choi,
K.-H.; Kim, H.-K.; Kim, D.-G.; Lee, D.-Y.; Nam, S.; Park,
J.-U. High-performance, transparent, and stretchable
electrodes using graphene-metal nanowire hybrid structures.
Nano Lett. 2013, 13, 2814-2821.
[56] Shigetomi S.; Ikari, T. Electrical and optical properties of n-
and p-InSe doped with Sn and As. J. Appl. Phys. 2003, 93,
2301-2303.
15
[57] Micocci, G.; Tepore, A.;, Rella, R.; Siciliano, P.
Investigation of deep levels in Zn-doped InSe single crystals.
J. Appl. Phys. 1992, 71, 2274-2279.
[58] Sánchez-Royo, J. F.; Pellicer-Porres, J.; Segura, A.;
Gilliland, S.-J.; Avila, J.; Asensio, M.-C.; Safonova, O.;
Izquierdo, M.; Chevy, A. Buildup and structure of the
InSe/Pt interface studied by angle-resolved photoemission
and x-ray absorption spectroscopy. Phys. Rev. B 2006, 73,
155308.
[59] Ning, J.; Xiao, G.; Wang, C.; Liu, B.; Zou, G.; Zou, B.
Synthesis of doped zinc blende-phase InSe:M (M = Fe and
Co) nanocrystals for diluted magnetic semiconductor
nanomaterials. Cryst. Eng. Comm. 2013, 15, 3734-3738.
[60] Chevy, A. Improvement of growth parameters for
Bridgman-grown InSe crystals. J. Cryst. Growth 1984, 67,
119-124.
[61] Hohenberg, P.; Kohn, W. Inhomogeneous electron gas.
Phys. Rev. 1964, 136, B864-B871.
[62] Kohn W.; Sham, L. J. Self-consistent equations including
exchange and correlation effects. Phys. Rev. 1965, 140,
A1133-A1138.
[63] Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera,
J.; Ordejón, P.; Sánchez-Portal, D. The SIESTA method for
ab initio order-N materials simulation. J. Phys: Condens.
Matter 2002, 14, 2745-2779.
[64] Artacho, E.; Anglada, E.; Diéguez, O.; Gale, J. D.; García,
A.; Junquera, J.; Martin, R. M.; Ordejón, P.; Pruneda, J. M.;
Sánchez-Portal, D.; Soler, J. M. The SIESTA method;
developments and applicability. J. Phys: Condens. Matter
2008, 20, 064208.
[65] Sánchez-Portal, D.; Ordejón, P.; Canadell, E. Computing
the properties of materials from first principles with SIESTA.
Structrure and Bonding 2004, 113, 103-170.
[66] Perdew, J. P.; Zunger, A. Self-interaction correction to
density-functional approximations for many-electron systems.
Phys. Rev. B 1981, 23, 5048-5079.
[67] Trouiller, N.; Martins, J. L. Efficient pseudopotentials for
plane-wave calculations. Phys. Rev. B 1991, 43, 1993-2006.
[68] Kleinman, L.; Bylander, D. M. Efficacious form for model
pseudopotentials. Phys. Rev. Lett. 1982, 48, 1425-1428.
[69] Louie, S. G.; Froyen, S.; Cohen, M. L. Nonlinear ionic
pseudopotentials in spin-density-functional calculations.
Phys. Rev. B 1982, 26, 1738-1742.
[70] Artacho, E.; Sánchez-Portal, D.; Ordejón, P.; García, A.;
Soler, J. M. Linear-scaling ab-initio calculations for large
and complex systems. Phys. Status Solidi B 1999, 215,
809-817.
[71] Monkhorst, H. J.; Pack, J. D. Special points for
Brillouin-zone integrations. Phys. Rev. B 1976, 13,
5188-5192.
16
Electronic Supplementary Material
Electronic structure, optical properties, and lattice dynamics in atomically thin Indium Selenide flakes
Juan F. Sánchez-Royo1 (), Guillermo Muñoz-Matutano1,†, Mauro Brotons-Gisbert1, Juan P. Martínez-Pastor1 (), Alfredo Segura1,2, Andrés Cantarero1, Rafael Mata1, Josep Canet-Ferrer1, Gerard Tobias3, Enric Canadell3, Jose Marqués-Hueso4, and Brian D. Gerardot4
1 ICMUV, Instituto de Ciencia de Materiales, Universidad de Valencia, P.O. Box 22085, 46071 Valencia, Spain. 2 MALTA-Consolider Team, Institut de Ciència dels Materials-Dpto. de Física Aplicada, Universitat de València, E-46100 Burjassot
(València), Spain. 3 Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus de la UAB, 08193 Bellaterra, Barcelona, Spain. 4 Institute of Photonics and Quantum Science, SUPA, Heriot-Watt University, Edinburgh EH14 4AS, UK.
† Present address: Optics and Quantum Communications group, ITEAM, UPV, Valencia, Spain.
Supporting information to DOI 10.1007/s12274-****-****-* (automatically inserted by the publisher)
1 Identification of thin InSe films by optical microscopy
Optical microscopy has been reported to be a simple, quantitative and universal way to locate and identify the
thickness of 2D materials such as graphene [3,72-74] and semiconducting transition metal dichalcogenides such
as MoS2, WSe2 or NbSe2 deposited on SiO2/Si wafers [75,76]. For a certain thickness of SiO2, it has been found
that even a single layer of these materials can give sufficient, albeit feeble, contrast to allow a successful
identification of a few micron-sized crystallites among thicker flakes scattered around them.
We have calculated the contrast of few InSe single layers exfoliated onto SiO2/Si substrates as a function of the
illumination wavelength and the SiO2 thickness, in order to determine the optimal imaging conditions for their
optical detection. In analogy with graphene and dichalcogenides [73-76], the contrast between the InSe flakes
and the underlying SiO2 substrate is due to a phase shift of the interference color and material opacity. In order
to calculate this contrast, we consider the stacking of two thin films (the InSe nanolayer and the SiO2 layer
underneath) on top of a third semi-infinite Si film (Supplementary Fig. S1). The InSe nanolayer is modeled by a
thin homogeneous film of thickness d1 = N x d, with N the number of InSe monolayers and d = 0.8 nm the
thickness of a single InSe single layer. For these calculations, flakes containing one-to-five InSe monolayers
were assumed to have the same optical properties, characterized by a real refractive index ñ1=2.5 [24,77]. The
SiO2 layer of thickness d2 is optically characterized by a wavelength-dependent real refractive index ñ2(λ) [78],
ranging from 1.47 at 400 nm to 1.455 at 700 nm. We have considered a semi-infinite silicon substrate
underneath, with a wavelength-dependent complex refractive index ñ3 [78]. Under normal light incidence, the
intensity of reflected light from the stacking of two thin films on top of a semi-infinite layer can be calculated
by the Fresnel equations [73,79]:
17
1 2 1 2 1 2 1 2
1 2 1 2 1 2 1 2
2( ) ( ) ( ) ( )
01 12 23 01 12 231 ( ) ( ) ( ) ( )
01 12 01 23 12 23
i i i i
i i i i
r e r e r e r r r eR n
e r r e r r e r r e
, S1
where
牋
i j
ij
i j
n nr
n n
are the relative indices of refraction of the different interfaces,
2 i ii
d n
are the phase
shift induced by changes in the optical path and ñ0, ñ1, ñ2 and
Supplementary Figure S1 Scheme of the geometry considered in the theoretical contrast calculations. It is composed by an InSe
nanolayer with thickness d1 and complex index of refraction ñ1. The nanolayer is deposited on a SiO2 layer characterized by thickness d2
and index of refraction ñ2 that is grown on top of a degenerately doped Si substrate characterized by an index of refraction ñ3.
ñ3 are the refractive indices of air, InSe, SiO2, and Si, respectively. On the other hand, the reflected light intensity
in the absence of an InSe flake can be found by substituting ñ1 = 1 in Equation S1:
2 2
2 2
2( ) ( )
02 231 ( ) ( )
02 23
1i i
i i
r e r eR n
e r r e
, S2
where 0 202
0?
牋
n nr
n n
is the relative index of refraction at the interface between air and the dielectric thin film.
Therefore, the contrast, as defined as the relative intensity of reflected light in the presence and absence of the
InSe flake, can be written as
1 1
1
1Contrast ?
1
R n R n
R n
. S3
Supplementary Figure S2 shows the theoretically calculated contrast of (a) a single InSe monolayer and (b)
three InSe single layers (one hexagonal-unit-cell thick) as a function of the incident light wavelength and SiO2
thickness. For both cases, the contrast for visible light wavelengths exhibits three characteristic bands with high,
positive contrast and three bands with high, negative contrast. As can be seen from this figure, these high
contrast bands are more intense in the case of the three InSe monolayers calculations. Despite of this, if the
18
excitation wavelength and the SiO2 thickness are properly chosen, the contrast of the single InSe monolayer is
large enough to detect it. Under green illumination and with the widely used 300 nm SiO2 substrates, it is
possible to obtain contrast values around 0.1, which is higher than the contrast of 0.08 obtained for a single
graphene monolayer using the same criteria [74].
Supplementary Figure S2 Color plot of calculated contrast as a function of incident light wavelength and SiO2 layer thickness for (a)
an InSe monolayer and (b) three InSe monolayers (one unit cell thick). Both plots have been drawn in the same contrast scale.
Supplementary Figure S3 gives the calculated contrast spectra of InSe samples with different thicknesses (from
one to five single layers) on SiO2(300 nm)/Si substrates under normal incident light based on the Fresnel
equation (Eqs. S1-S3). As can be seen from this Figure, the absolute contrast value increased as the number of
layers increased, which made unambiguous identification of the thickness of InSe possible.
Supplementary Figure S3 Contrast spectra of InSe flakes with different thicknesses (one, two, three, four and five monolayers) on
SiO2(300 nm)/Si substrate calculated under normal incidence based on the Fresnel equation.
19
2 Photoluminescence in bulk InSe
In the PL spectrum recorded in the bulk sample (Supplementary Fig. S4) on a backscattering geometry
(standard lens, no confocal), we observe several contributions that we split in two groups:
Supplementary Figure S4 PL spectrum measured in a bulk InSe sample. Deconvolution of this spectrum has resulted into five
Gaussian components, which have been included.
In the first group, we consider three high energy and narrow peaks (3-9 meV) at 1.337, 1.329 and 1.319 eV.
The origin of these PL lines in either not purposely or slightly doped InSe has not been a subject of intensive
research and the authors that studied this semiconductor attributed them to free and different donor bound
excitons [45,47,51,80]. On the other hand, native donors in not purposely doped InSe (always n-type) are due
to In in excess that can be give rise to shallow donor levels if incorporated in interstitial sites of the bulk
lattice, but also to deep levels when they are presumably located in intralayer sites [50]; in fact most of these
impurities can be promoted to interstitial sites (shallow donors) after annealing. The ionization energy of
shallow donor levels was found to be 24 meV and hence the line at 1.319 eV (1.35 eV would be the InSe
bandgap energy at low temperatures) [26] can be reasonably assigned to the h-D0 and D+X (D0 and D+
associated to shallow donors) transitions.
In a second group of peaks, we detect a relatively wide PL band (21 meV wide) at 1.310 meV that could be
related to h-D0 transitions, either shallow (because of its overlapping with the narrower line at 1.319 eV) and
deeper donor levels at inter-layer sites. Given the random nature of the spatial position for these inter-layer
impurities, the donor binding energy can vary over a wide range of values. Such energy distribution of
donor levels can be responsible of this PL band at 1.310 eV and its low energy tail (simulated by using a wide
Gaussian band at around 1.288 eV).
20
3 Micro-Photoluminescence in thick InSe flakes.
The -PL spectra showed below were acquired in a bulk-like (12 nm) InSe flake, as a function of excitation
power (Supplementary Fig. S5(a)) and temperature (Supplementary Fig. S6(a)). The -PL spectra showed in
the Supplementary Fig. S5(a) are characterized by recombination times comprised between 5 and 9 ns
(Supplementary Fig. S5(b)) depending on detection wavelength where the PL transient is measured. By
increasing power, the high energy PL tail grows in importance. In addition to this, the recombination time
and the -PL intensity decrease as temperature increases (Supplementary Fig. S6(a)-S6(b)). All these facts
point out to the impurity-related origin of the -PL signal measured in the flakes at low temperatures.
Supplementary Figure S5 (a) Low-temperature -PL spectra measured in a bulk-like ~12 nm thick InSe flake, at two different
excitation-power values. (b) -PL decay curves obtained at three different energy positions in the -PL spectrum of (a) measured with
P=23.8 W. Decay curve at the top (bottom) was measured at the low (high) energy side of the -PL spectrum. Decay times are
indicated
Supplementary Figure S6 (a) -PL spectra and (b) -PL decay curves measured in a bulk-like ~12 nm thick InSe flake with P=23.8
W, at different temperatures. Decay times are indicated
21
4 Micro-Photoluminescence in single-terraced InSe flakes.
The -PL spectra showed below were acquired in bulk-like InSe flakes (1-and 4-labelled flakes) and in
single-terraced, 8 nm-thick, flakes (2- and 5-labelled flakes). The -PL spectra of the bulk-like flakes are very
similar to that measured in the thickest terraces of the flake showed in the Fig. 3 of the main text. The line shape
of the -PL spectra measured in the 8 nm-thick flakes seems to depend on the flake studied: The -PL spectra
acquired in the 2- and 5-flakes show a wider or narrower lineshape, respectively, than that obtained for the
d-labelled terrace (8 nm-thick) of the flake showed in Fig. 3. In all cases, deconvolution of the -PL spectrum
can be performed by taking into account some of the six P1-P6 components (see main text).
Supplementary Figure S7 Left plots are optical images of single-terraced InSe flakes. Thickness of the selected 1-5 flakes was
determined by AFM. Right plots show the -PL spectra recorded in the 2- and 5-flakes, 8 nm-thick, and in bulk-like terraces joined to
these thin flakes. The results of the deconvolution of the -PL spectrum measured in the 8 nm-thick flakes are shown. Note that the
peaks obtained from this deconvolution have been represented using the same color-code than that used in Fig. 3.
22
Taking into account these results, the following facts can be remarked:
i) The presence of the P1-P6 components is a common signature of all the low-temperature -PL spectra
recorded in the thin InSe flakes, which supports the physical origin of each one of the P1-P6 components
discussed in the main text.
ii) The -PL spectra obtained in terraces with similar thickness can be deconvoluted into P1-P6 components
whose relative intensities appear to differ from flake to flake. This fact can be understood on the basis of the
hypothesis stated in the main text on the unidirectional diffusion of photo-generated carriers from thicker to
thinner flakes. This hyphothesis conveys that -PL spectrum lineshape can be strongly determined by the
characteristics (i.e., thicknesses variations) of the flake edge. In fact, the fact 5-flake has a sharper flake-edge
than the 2-flake, as it can be appreciated in the images of the Supplementary Fig. S7, which would explain the
differences observed between the lineshapes of the -PL spectra recorded in these flakes.
Finally, it must be noticed that no -PL signal was recorded when very thin flakes of uniform thickness were
probed, as the 3-labeled flake (3 nm-thick) showed in the Supplementary Fig. S7. This behavior contrasts to that
observed in terraces of similar thicknesses which are part of multi-terraced flakes (as the f-labeled terrace of the
flake of Fig. 3). At this point, it is worth to mention that this observation applies to very thin flakes (already
5-nm thick single-terraced flakes gave rise to -PL signal, when probed). We believe that this observation is a
clear consequence of the unidirectional diffusion of photo-generated carriers in multi-terraced flakes: In very
thin flakes of uniform thickness, photogenerated carriers isotropically diffuse before recombination. However,
in multiterraced flakes, these carriers would tend to diffuse towards the flake edge. Comparing these two
situations, surface recombination would tend to play a more determinant role in flakes of uniform thickness.
Therefore, the -PL signal intensity coming from these uniform flakes would suffer from a strong attenuation
due to the fact that surface recombination is mainly determined by non-radiative processes, as surface defects
or SiO2/InSe interface traps.
Supplementary References
[72] Bassani, F.; Tosatti, E. Phys. Lett. A 1968, 27, 446.
[73] Blake, P. et al. Appl. Phys. Lett. 2007, 91, 063124.
[74] Wang, Y Y. et al. Nanotechnology 2012, 23, 495713.
[75] Benameur, M. M. et al. Nanotechnology 2011, 22, 125706.
[76] Castellanos-Gomez, A.; Agraït, N.; Rubio-Bollinger, G. Appl. Phys. Lett. 2010, 96, 213116.
[77] Viswanathan, C.; Rusu, G. G.; Gopal, S.; Mangalaraj, D.; Narayandass, Sa. K. J. Optoelectron. Adv. M. 2005, 7, 705.
[78] Borghesi, A.; Guizzetti, G. Handbook of Optical Constants of Solids II, 449–460 (Academic, Boston, 1991).
[79] Anders, H. Thin Films in Optics (London: Focal) (1967).
[80] Zhirko, Y.; Skubenko, N.; Dubinko, V.; Kovalyuk, Z.; Sydor, O. J. Mat. Sci. and Eng. A 2013, 3, 162.
23
————————————
Address correspondence to Juan F. Sánchez-Royo, [email protected]; Juan P. Martínez-Pastor, [email protected]