1
Titanium-Carbide MXenes for Work Function and Interface Engineering in
Perovskite Solar Cells
A. Agresti(a,b)§, A. Pazniak(c)§, S. Pescetelli(a)§, A. Di Vito(a), D. Rossi(a), A. Pecchia(d), M. Auf der
Maur(a), A. Liedl(e), R. Larciprete(e,f), Denis V. Kuznetsov(c), D. Saranin(b), A. Di Carlo(a,b)*
a) CHOSE - Centre for Hybrid and Organic Solar Energy, Department of Electronic Engineering,
University of Rome Tor Vergata, via del Politecnico 1, 00133, Rome, Italy;
b) LASE – Laboratory of Advanced Solar Energy, National University of Science and Technology
“MISiS”, Leninsky prospect 4, 119049, Moscow, Russia.
c) Department of Functional Nanosystems and High-Temperature Materials National University of
Science and Technology “MISiS”, Leninsky prospect 4, 119049, Moscow, Russia
d) Istituto per lo Studio Materiali Nanostrutturati - CNR, Via Salaria km 29.600, 00014, Rome, Italy.
e) INFN-LNF, P.O. box 13, 00044 Frascati (Rome) Italy.
f) CNR-Institute for Complex Systems (ISC), Via dei Taurini 19, 00185 Rome, Italy
§: The authors contributed equally to this work
*corresponding author: aldo.dicarlo@uniroma2.it
ABSTRACT
In order to improve the efficiency of perovskite solar cells (PSCs), careful device design and tailored interface
engineering are needed to enhance optoelectronic properties and the charge extraction process at the
selective electrodes. Here, we use two-dimensional transition metal carbides (the MXene Ti
3
C
2
T
X
) with
various termination groups (T
X
) to tune the work function (WF) of the perovskite absorber and the TiO
2
electron transport layer (ETL), and to engineer the perovskite/ETL interface. Ultraviolet photoemission
spectroscopy measurements and Density Functional Theory calculations show that the addition of Ti
3
C
2
T
X
to
halide perovskite and TiO
2
layers permits to tune the materials’ WFs, without affecting other electronic
properties. Moreover, the dipole induced by the Ti
3
C
2
T
X
at the perovskite/ETL interface can be used to change
the band alignment between these layers. The combined action of WF tuning and interface engineering can
lead to substantial performance improvements in MXene-modified PSCs, as shown by the 26% increase of
power conversion efficiency and hysteresis reduction with respect to reference cells without Mxene.
The rapid development of perovskite solar cells (PSCs) carried out in the last decade demonstrated the
potentiality of this PV technology to compete on equal footing with traditional inorganic PV or to work in
synergy with established silicon technology in tandem cell configuration.
1
The remarkable efforts devoted to
optimize the PSCs in term of device architecture
2
, perovskite composition
3
and charge collecting electrodes
4
produced a large wallet of emerging materials comprising organic polymers,
5
oxides,
6
two-dimensional (2D)
materials
7
and others
8
able to tune the device properties
9
and eventually to boost their performance and
stability.
10
Usually, a typical PSC is composed of a perovskite active layer, sandwiched between two selective
charge transport layers (CTLs) and electrodes for negative and positive charge extraction. These are selected
carefully by taking into account their compatibility with the underneath layers,
11
their charge mobility
12
and
the energy level alignment with the perovskite absorber.
13
However, being PSCs composed of several layers,
interfaces ultimately play a crucial role in ruling device performance and stability. Charge transfer at the
interfaces,
14
interface band alignment,
15
interfacial vacancies,
16
defects due to poor adhesion between
layers
17
and energy barriers have a strong impact on electrical device characteristics. From here, the pivotal
role that interface engineering (IE)
18
has recently gained in the PSCs field.
19
In this regard, 2D materials can
be inserted as inter
20
or buffer layers
21
by modifying the chemical/physical properties of involved layers at
2
the interface and eventually to improve the charge injection/collection at perovskite/CTLs interfaces. One of
the main potentialities of 2D materials is the possibility to easily tailor their electronic structure, such as work
function (WF)
22,23
or band gap, by proper functionalization
24
or by quantum confinement.
25
Fine-tuning of the
WF allows to obtain appropriate energy level alignment leading to an ideal energy offset between perovskite
active layer and CTLs, eventually inducing a built-in potential for efficient charge collection at the electrodes.
Recently, a new family of 2D transition metal carbides, nitrides and carbonitrides (MXenes)
26
with a general
formula M
n+1
X
n
T
x
(n = 1, 2, 3), where M represents an early transition metal, X is a carbon and/or nitrogen
atom and T
X
the surface-terminating functional groups came out as a promising class of 2D materials in many
applications
26,27
owing to their physical and chemical properties.
26,28–30
Rich chemistry and surface
termination make MXenes unique 2D materials with huge possibilities to tune their electronic properties. In
fact, MXenes offer the possibility to vary the WF by choosing the proper transition metal as well as the X
element.
31
Moreover, during the synthesis of MXenes, their surfaces are naturally functionalized, which
changes the electrostatic potential near the surfaces and affects the electronic structure, significantly shifting
the WF.
32
As density functional theory (DFT) predicts, surface termination strongly influences the density of
states
33
and the WF of MXenes
34
which can range from 1.6 eV (for OH-termination) to 6.25 eV (for O-
termination).
34,35
This opens new opportunities for MXenes applications in optoelectronics and in particular
in photovoltaics where already some initial studies have been presented for organic solar cells
36
, Si solar
cells,
37
dye-synthesized solar cells
38
and PSCs.
39,
40
In the PSC case, Ti
3
C
2
T
x
MXenes have been incorporated
into the perovskite absorber showing an improved morphology and an enhanced PCE (+12%) with respect to
the reference cell without MXenes
39
or into the SnO
2
electron transporting layer (ETL)
40
to provide superior
charge transfer paths that permits to enhance the PCE (+6.5%) with respect to the reference cells.
In order to reveal the role of MXenes in perovskites and provide a clear strategy on the use of MXenes for
WF and IE in PSCs, we synthesized Ti
3
C
2
T
x
MXene and performed an extensive characterization of the MXene
structure and electronic properties (Figure 1).
Figure 1: Characterization of Ti
3
C
2
T
x
MXene. a, Schematic structure of Ti
3
C
2
T
x
MXene. Surface terminations (T
x
) are a
mixture of F, O, and OH. b, TEM image of Ti
3
C
2
T
x
MXene flakes. The corresponding selected area electron diffraction
(SAED) pattern is reported in the inset. UPS spectra measured with photon energy of 40.81 eV on the MXene flakes and
FTO substrate supporting them are reported in left panel c and right panel d showing secondary electron cut-off and
valence band region, respectively.
3
After chemical etching and exfoliation by sonication, single-layer Ti
3
C
2
T
x
nanosheet (Figure 1a) consists of
two carbon atoms which bind three titanium ones as elementary units. The outer Ti layers tend to terminate
with functional groups (Tx) such as O, OH, and F (Figure 1a), which are randomly distributed on the MXene
surface. The MXene structure is characterized by the presence of single atomically-thin transparent flakes of
2D titanium carbides with hexagonal symmetry which follows the parent MAX phase after exfoliation (Figure
1b). Ti
3
C
2
T
x
sheets have irregular edges and size distribution ranging generally from 1.5 to 2.5 µm. Details
about MXene composition were extracted by XPS measurements reported in Figures S1 and S2a. In
particular, the ratio between the F : OH : O functional groups is estimated to be 1.6 : 0.65 : 0.34. Based on
these results, it can be concluded that the MXene surface randomly ends with the F, OH and O groups, with
a prevalence of fluorine functional groups.
The electronic properties of Ti
3
C
2
T
x
was featured by WF measurement using ultraviolet photoelectron
spectroscopy (UPS). We found that the WF of synthesized MXene flakes deposited on FTO glass is rather low
and equal to 3.7 eV (Figure 1c). The spectra measured in the valence band region (Figure 1d) show that the
intensity extends up to the Fermi level and severely increases for binding energy (BE) close to 4 eV, likely due
to the contribution of the oxidized phases detected in the XPS spectra (see Figure S1a,b). In general, the
reported UPS measured WF values for Ti
3
C
2
T
x
MXene range from 3.4 eV
41
to 4.62 eV.
42
Such large modulation
of the WF could be attributed to the different etching environment dramatically affecting the surface
termination. The WF value of synthesized MXene is in agreement with the XPS results and should be due to
the substantial percentage of hydroxyl groups terminating the Ti
3
C
2
T
x
matrix. This finding is supported by
theoretical calculations
32
(see also the discussion in SI) and experimental observations reported in
literature.
41
The low WF of MXene flakes could be exploited to finely control the energy level alignment between the
perovskite absorber layer and the CTLs. To this end, we probed the effect of the MXene employed as additive
in perovskite and TiO
2
layers onto their WFs by combining UPS (Figure 2) and XPS (Figure S2b)
measurements.
43
The WF of pristine perovskite, determined from the secondary electron onset, is 4.72 eV,
which is shifted to 4.37 eV after the addition of MXene (see Figure 2a), i.e. 0.35 eV lower than that of the
undoped perovskite. The valence band spectrum measured on the MXene-doped perovskite film and the
energy gap appear quite like the pristine ones. (Figure 2b and S3). The energy diagram in Figure 2c
schematizes the WF and Ionization Energy (IE) reduction of MXene-doped perovskite.
Figure 2: UPS curves of pristine and MXene-doped perovskite films. a, UPS spectra around the secondary electron
cut-off. b, UPS spectra in the valence band region. For the pristine perovskite, the valence band maximum (VBM),
determined by the intercept to zero of the intensity plotted in logarithmic scale (see inset in panel b) is at 1.46 eV below
the Fermi level, in good agreement with previous findings.
43
c, Energy scheme for undoped and MXene-doped
perovskite with respect to the Fermi level.
4
We exclude that in perovskite deposition the addition of MXene can change the perovskite crystal growth
and the resulting morphology since no significant changes have been observed in the scanning electron
microscopy (SEM) images reported in SI (Figure S4).
UPS characterizations were performed also for the TiO
2
ETL with and without addition of MXene (Figure S5).
In this case, MXene doping leaves the shape and the maximum of the valence band almost unchanged. At
the same time, the WF slightly decreases from 3.91 eV for the TiO
2
sample to 3.85 eV in the case of MXene-
doped TiO
2
. The small WF differences between the pristine and doped TiO
2
layers compared to the WF shift
induced by MXene in perovskite can be related to the oxidation of Ti
3
C
2
T
x
MXene (Figure S6). However, the
WF decrease observed in the doped TiO
2
layer suggests that MXene does not oxidize completely when
annealed with TiO
2
as confirmed by the XRD pattern measured for sintered MXene-doped TiO
2
(Figure S7).
In order to understand the origin of the perovskite WF change induced by the interaction with MXene, we
performed DFT calculations of the perovskite/Ti
3
C
2
T
X
with OH and O termination. To make this calculation
feasible, we consider the single cation CH
3
NH
3
PbI
3
(MAPbI
3
) perovskite having experimentally verified that
MXene doping induces a WF shift also for this perovskite (Figure S8). Calculation details are reported in SI
(see Figures S9 and S10). DFT calculations reported in Figure 3 show that the charge transfer at the
MAPbI
3
/Ti
3
C
2
(OH)
2
interface induces the formation of an interface dipole, causing an important reduction of
the WF and affecting the band alignment of the system. The transferred charge density is localized at the
very first Pb-I layer, therefore involving less than 1 nm of the surface. The formation of an interface dipole
can also be deduced from the vacuum potential slopes observed in Figure 3a and 3b. In the case of OH
terminated MXene, the slope is higher than O termination, implying a larger interface dipole. Moreover, in
MAPbI
3
/Ti
3
C
2
O
2
the potential slope (i.e. the interface dipole) has the opposite direction with respect to OH
termination case. Figure 3c and 3d show the projected bands of the MAPbI
3
/MXene slabs, where the red
colour represents the contribution from the inner atoms of the MAPbI
3
slab, falling in the grey boxes drawn
in Figure 3a and 3b.
5
Figure 3: DFT calculation of the MAPbI
3
/MXene structure. a, b Electrostatic potential averaged over planes
perpendicular to the MAPbI
3
/Ti
3
C
2
(OH)
2
and MAPbI
3
/Ti
3
C
2
O
2
interface, respectively. The computed structures are
shown within the plots, where green, magenta, blue, yellow, cyan, grey and red spheres represent I, Pb, N, C, H, Ti and
O atoms, respectively. The red dashed lines represent the dipole corrected vacuum levels. The Fermi energy is set to
zero, so that the vacuum potential, just away from the MXene surface, corresponds to the WF, of the MAPbI
3
/Ti
3
C
2
T
x
interface, as depicted in the panels. We can see that the WF derived for the OH terminated MXene configuration,
WF
OH
≅2.1 eV, is substantially smaller than the value obtained for the O terminated structure, where WF
O
≅5.7 eV.
Notably, similar calculations for the F terminated MXene do not show a significant variation of MAPbI
3
WF. c, d,
Projected band structures of the MAPbI
3
/MXene slabs for OH and O termination of the MXene, respectively.
Contribution from the bulk part of the MAPbI
3
slab (grey box in panels a, b) are coloured in red. The valence band edge
is set to zero. The bulk band gap of the MAPbI
3
is indicated by the shaded area.
A remarkable outcome emphasised in Figures 3c and 3d is that, even in such small structures, the MAPbI
3
bandgap is barely affected by the interaction with MXene, remaining at 1.7 eV. This gap independence is a
feature of MXene additives also seen in experiments. We conclude that the interaction is mainly electrostatic,
owing to the dipole formation at the interface. A similar result is obtained when considering a symmetric