Benoit Galvani1, Daniel Suchet2, Amaury Delamarre3, Marc Bescond4, Fabienne Velia Michelini1, Michel Lannoo1, Jean-Francois Guillemoles5, Nicolas Cavassilas1. 1. Aix Marseille Université, CNRS, Université de Toulon, IM2NP UMR 7334, 13397 Marseille, France. 2. Ecole Polytechnique, IPVF, Institut Photovoltaque d' Ile-de-France, 30 RD 128, 91120 Palaiseau, France. 3. Centre de Nanosciences et de Nanotechnologies, 91120 Palaiseau, Ile-de-France, France. 4. LIMMS, CNRS-Institute of Industrial Science, UMI 2820, University of Tokyo, 153-8505 Tokyo, Japan. 5. IPVF, Institut Photovoltaque d' Ile-de-France, 30 RD 128, 91120 Palaiseau, France.
Abstract
We numerically investigate the impact of electron-phonon scattering on the optical properties of a perovskite material (CH3NH3PbI3). Using nonequilibrium Green function formalism, we calculate the local density of states for several values of the electron-phonon scattering strength. We report an Urbach-like penetration of the density of states in the band gap due to scattering. A physical analytical model allows us to attribute this behavior to a multiphonon process. Values of Urbach energy up to 9.5 meV are obtained, meaning that scattering contribution to the total experimental Urbach energy of 15 meV is quite important. We also show that the open-circuit voltage V oc, for a solar cell assuming such a material as an absorber, depends on the scattering strength. V oc loss increases with the scattering strength, up to 41 mV. Finally, an unexpected result of this study, is that the impact of electron-phonon scattering on Urbach tail and V oc increases with the phonon energy. This low value in perovskite (8 meV) is therefore an advantage for photovoltaic applications.
We numerically investigate the impact of electron-phonon scattering on the optical properties of a perovskite material (CH3NH3PbI3). Using nonequilibrium Green function formalism, we calculate the local density of states for several values of the electron-phonon scattering strength. We report an Urbach-like penetration of the density of states in the band gap due to scattering. A physical analytical model allows us to attribute this behavior to a multiphonon process. Values of Urbach energy up to 9.5 meV are obtained, meaning that scattering contribution to the total experimental Urbach energy of 15 meV is quite important. We also show that the open-circuit voltage V oc, for a solar cell assuming such a material as an absorber, depends on the scattering strength. V oc loss increases with the scattering strength, up to 41 mV. Finally, an unexpected result of this study, is that the impact of electron-phonon scattering on Urbach tail and V oc increases with the phonon energy. This low value in perovskite (8 meV) is therefore an advantage for photovoltaic applications.
Perovskite materials are widely studied
for both device conception
and theoretical understanding. Among the variety of perovskites materials,
organic–inorganic hybrid perovskites are the most promising
materials for optoelectronic applications. First, photovoltaic cells
made of hybrid perovskites lacked stability and had a low solar conversion
efficiency, of the order 3%. Nowadays, fabrication techniques of perovskites
have been largely improved, making them more stable and reliable to
use in solar cells. Production costs have also been drastically reduced,
paving the way to further decrease the PV electricity price.[1] This fast evolution of technology has led to
an increase of efficiency, up to 24% in less than 10 years,[2] and have reached over 28% in Si-perovskite tandem
devices.[3]Hybrid perovskite represents
one of the most promising materials
for reliable solar conversion technology, thanks to their excellent
optical and electrical properties. They exhibit high carrier diffusion
lengths, reaching up to 1 μm even in nanostructures.[4] High carrier lifetime has also been measured
to be around several microseconds.[5] Moreover,
it has been shown that hybrid perovskites have remarkable photon absorption
properties.[6] These properties have also
been observed in perovskite layers as thin as 50 nm.[7] The main advantage of such thin films is to be integrated
in lightweight, flexible, and transparent solar cells[8] that could find applications in wearable electronics.However, remarkable properties of these materials cannot be fully
understood considering the peculiar structure of hybrid perovskites.
On the one hand, with these excellent carrier-transport properties,
we could expect hybrid perovskites to exhibit high carrier mobilities.
Yet, several groups have reported low values for carrier mobilities
in perovskite-based devices. Mei et al. found mobilities of 1 cm2/(V s) in a field-effect transistor.[9] Shi et al. estimated the electron mobility in a single crystal to
be 115 cm2/(V s)[10] Microwave
conductivity measurements gave carrier mobilities from 6 to 60 cm2/(V s).[11,12] These measured values are quite
low compared to mobilities in other inorganic materials used in PV
cells (electron mobility is equal to 1500 cm2/(V s) in
silicon) and could indicate some disorder in the crystalline structure.
On the other hand, it is possible to extract information about the
quality of the material from the absorption features below the band
gap. The Urbach rule gives the absorption coefficient below the band
gap as a function of photon energy and temperature. It is described
by a “tail” with the expressionwhere α0 is a constant, Eg is the material band gap, hν is the incident photon energy, and EU is the Urbach energy, corresponding to the width of the exponential
part of the absorption below the band gap. This empirical formula
is an important tool to study the optical absorption due to its simplicity
and applicability for various types of materials and structures. This
phenomenon has three principal physical causes which are the crystal
disordering,[13] trap states due to impurities[14] and electron–phonon scattering (EPS).[15,16] Depending on the type of material, one of these causes dominates.
In amorphous material, the Urbach energy, which can reach up to 50
meV,[17] is mainly due to disordering, while
in well-ordered materials like GaAs, the Urbach energy, which is as
low as 7.5 meV,[18,19] is mainly due to the EPS. In
the case of hybrid perovskite-based device, the Urbach energy has
been estimated experimentally to be 15 meV.[20] However, the origin of this band tail is still debated in the literature.
In ref (21), the authors
attribute the latter to the presence of trap states in the band gap.
The understanding of this behavior is important since it modifies
the characteristic of the corresponding solar cell with a Stokes shift
and impacts the open-circuit voltage oc.[22] The definition of the optical band gap can also be discussed.[23,24]In hybrid perovskites, the scattering with polar optical phonons
is expected to be large since (i) the difference between ϵ∞ and ϵstat (the dynamic and the static
dielectric constant respectively) is large[25] and (ii) the small phonon energy implies a large number of phonons
(all parameters are reported in Table ). Moreover, rotations of the CH3NH3 organic matrix may play a key role in quantum transport properties.[26] Random rotations of cations induce fluctuations
of the electrostatic potential. Electrons and holes in the material
become localized and spatially separated, affecting their transport
properties and the corresponding mobilities. It can then be expected
that EPS also impacts the Urbach tail.
Table 1
Constant Parameters of CH3NH3PbI3
parameter
value
me*/m0
0.23
mh*/m0
0.17
ϵ∞
6.0
ϵstat
29.0
Epop (meV)
8.0
Numerical study of Urbach tail assumes that the acoustic
phonon
has already been conducted.[16] Recently,
using nonequilibrium Green’s functions (NEGF) formalism,[27] impurities and EPS have been considered to allow
to reproduce the band tail and the band gap narrowing in a nanowire.
Using the same formalism,[28] in the present
work, we numerically study the impact of the polar optical phonon
on the optical and electrical properties of hybrid perovskite materials.
We show that such an interaction modifies the electronic density of
states (DOS), leading to the enhancement of an Urbach tail in the
band gap. To better interpret our results, we develop an original
analytical model of the DOS in the band gap. This model is in good
agreement with the numerical results assuming only well-known material
parameters such as effective masses, electron–phonon coupling,
and phonon energy. It also allows to illustrate that the Urbach tail
due to polar optical phonon is related to a multiphonon process. Finally,
calculations of the absorption and the emission current contributions
allow to estimate the variation of oc with respect to the EPS strength.
Methodology
To study the impact of EPS, we made numerical
calculations using
NEGF formalism within the self-consistent Born approximation.[29] This model is detailed in ref (28), but we recall here the
main elements needed for an easier understanding. We consider a three-dimensional
(3D) device, with wide transverse dimensions perpendicular to the
transport axis. This allows to assume a one-dimensional (1D) model
in the longitudinal direction for each transverse mode corresponding
to a propagation in the transverse plane with a given transverse wave
vector. In our case, we consider N = 30 transverse modes. For each mode, we define a Hamiltonian H, in the 1D site basis using the effective mass approximation, which
is suitable for the study of most crystalline material, including
perovskite.[30] This allows to calculate
the corresponding lesser G<(E, k) and
greater G>(E, k) Green’s functions,
relative to electron and hole densities, respectively. The DOS in
both conduction and valence bands is given by the retarded Green function GR(E, k). We consider the diagonal approximation
to describe the EPS and assume the effective mass approximation for
both conduction and heavy-hole valence bands. To make the calculations
with such formalism, we need to take the band diagram of the device
as well as the EPS strength and phonon density as input data in our
code. For calculations in a solar cell, the band diagram, relative
to electric field and band offset, is obtained by the Poisson–Schrödinger
self-consistent approach. To do so, we implemented the structure in
SCAPS software.[31] SCAPS is a solar cell
simulation program that calculates the optical and electronic characteristics
of multilayer semiconductor devices based on a drift-diffusion model.
In our case, we have used SCAPS to extract data relative to the band
diagram of our device as an input file of our Green’s function
simulations.In such a polar material, polar optical phonons
dominate the carrier
transport. While the model of Sarangapani et al. includes scatterings
with both polar optical phonon and charged impurities, and Grein et
al. assumed only acoustic phonons, we then only assume the polar optical
phonons. The strength of this electron–phonon interaction is
given bywhere S is the surface of
the transverse cross section of radius R = 30 nm; Epop is the polar optical phonon energy of the
material; ϵ0 is the vacuum permittivity; and ϵ∞ and ϵstat are the relative dynamic
and static permittivities, respectively. All used parameters are given
in Table . CEPS is a numerical fitting factor, which is
adjusted to a value such that the model accounts for the correct value
of the electron mobility.[32] The relation
between CEPS and mobility is shown in Table . Electron mobility
in hybrid perovskites has been experimentally measured from 1 to 100
cm2/(V s). Mechanisms controlling the carrier mobility
in this material are not clearly understood, since the CH3NH3 molecular rotations could have an impact on carrier-transport
properties.[26] From a formalism point of
view, these rotations are close to the electron–phonon interaction,
but with unknown parameters. In the following, we perform calculations
with 4 different values of CEPS corresponding
to mobilities ranging from 0.13 to 86 cm2/(V s) as reported
in the literature.
Table 2
Numerical Values of the Scattering
Strength Multiplicative Coefficient, with the Estimated Electron Mobility
and Calculated Urbach Energy
CEPS
μe– (cm2/(V s))
EU (meV)
1.6
86
5.25
5.0
18
6.6
10.0
1.7
7.9
20.0
0.13
9.5
The number of phonons per unit cell is given by the Bose–Einstein distributionwhere E = Epop = 8 meV and T = 300 K in the case
of phonons in CH3NH3PbI3. Since Epop is small in the considered perovskite material
(compared to kBT = 25.7
meV and compared for example to the GaAs value of 35 meV), is large. We consider the EPS by calculating
the corresponding interaction self-energieswhere M(k, k′) is the intermode coupling matrix, which depends on
both considered modes k and k′ and on Mpop.[33]Equation describes the interaction with electrons,
while eq accounts for
those with holes. Terms in correspond to phonon absorption, while
terms in correspond to phonon emission. Calculations
of self-energies are self-consistent, since G< and G> also depend on these
self-energies[28]NEGF calculations
allow us to estimate the current generated by
the cell in the dark as well as under illumination. From current calculations,
we can thus deduce the relation between Voc and EPS.The electrical characteristic of a solar cell is
close to the diode
characteristic. In the dark, the current generated by the cell is
given bywhere I0 is the
current relative to recombination of majority carriers in the cell
for a bias V = 0 V. In the dark, I0 compensates the minority carrier current and the total
current is zero. Under illumination, the current in the cell at a
bias V can be estimated bywhere Isc is the
short-circuit current of the cell, which is the photocurrent at zero
bias. Here, we consider that the photocurrent does not depend on the
bias V. It is then possible to estimate the open-circuit
voltage oc only with
the Green’s functions calculated at V = 0
Voc is calculated for each value of CEPS, and each oc is compared
to the ballistic counterpart (CEPS = 0)
to estimate δoc, the losses due to EPS. Isc and I0 are calculated usingandwhere is a band coupling parameter,[34]E is the electron energy, hν is the photon energy, and Ω is the solid
angle of the light source, which is equal to 6.79 × 10–5sr for Isc and π sr for I0. is the Bose–Einstein distribution
of photons (see eq )
at the energy hν and temperature Tsun = 6000 K and ρ(hν) iss
the photon DOS per unit energy and unit volumeI0 depends on G<(E + hν) and G>(E),
which are the electron density at energy E + hν and the hole density at energy E, respectively. An electron in the conduction band can recombine
with a hole in the valence band, emitting a photon of energy hν. On the other hand, eq shows that Isc depends on G<(E)
and G>(E + hν), which are the electron density at energy E and the hole density at energy E + hν. An electron in the valence band can interact with
a photon
with energy hν to populate a state in the conduction
band.
Results and Discussion
Density of States
Figure shows the local DOS calculated in a solar
cell with an hybrid perovskite active region. We performed the calculations
with a CEPS corresponding to an electron
mobility of 86 cm2/(V s) and in the ballistic approximation
(CEPS = 0). The two situations are compared
in Figure . In the
ballistic case, quantum tunneling leads to a small but finite penetration
of the DOS in the band gap. This DOS distribution has been shown to
impact the optical properties of ultrathin solar cells.[35] By comparison, the presence of EPS strongly
enforces the penetration of DOS within the band gap, which largely
dominates the one induced by quantum tunneling. The presence of these
near-band-edge states allow both electrons and holes to be either
photogenerated or recombined at energies lower than the material band
gap of 1.55 eV. To illustrate the evolution of broadening with the
interaction strength, we have also made local DOS calculations for
each value of the scattering strength. Figure shows these DOS in the middle of the device
versus the electron energy. The DOS tends to penetrate deeper in the
band gap as the EPS becomes stronger.
Figure 1
Local density of states in a perovskite-based
device (shown above)
as a function of position and electron energy, calculated (a) with
electron–phonon interaction and (b) without electron–phonon
interaction.
Figure 2
Logarithm of the density of states as a function of electron
energy
at a fixed position in the middle of the active region of the perovskite-based
device. Each color corresponds to a different scattering coupling
strength, and thus to a different value of electron mobility.
Local density of states in a perovskite-based
device (shown above)
as a function of position and electron energy, calculated (a) with
electron–phonon interaction and (b) without electron–phonon
interaction.Logarithm of the density of states as a function of electron
energy
at a fixed position in the middle of the active region of the perovskite-based
device. Each color corresponds to a different scattering coupling
strength, and thus to a different value of electron mobility.To get some physical insight, we developed an analytic
model showing
the relative importance of the various parameters. The gap density
of states is related to the imaginary part of the retarded Green’s
functionwhere E is the energy of the band edge
(we consider here the conduction band) and . Using eqs and 5 and
the approach developed in ref (33), we can write Γ in a continuous formwhich is valid for small electron concentrations.
As shown in the Supporting Information (eqs S5 and S9), this set of equations can be
reduced towith .This is a set of ladder equations,
which are related to multiphonon
process, leading to an exponential decrease of large negative energies.
In this regime, |E| ≫ Epop and eq become a simple differential equation, valid at large . To solve it, we use the expected exponential
behavior by writing D(E) = A exp(σ(E)) and incorporate
it into eq together
with a first-order expansion σ(E ± Epop) = σ(E) ± σ′(E)Epop. With this, all terms
in σ(E) disappear and, by expanding exp(±σ′(E)Epop) to second order and
keeping the dominant term in |E|, one getswhose integration givesThis depends upon two basic parameters Epop and E. Note that it depends upon 1/M linearly. This analytical
expression is
compared to the full numerical result in Figures and 4, showing an
excellent agreement at large energies, apart from a shift due to the
constant. In particular, we confirm that the slope of the Urbach tail
is proportional to Epop and also to EM3/4, which is linear in M and one that can relate to the mobility.
Figure 3
Density
of states in the considered device as a function of the
electron energy. Each color corresponds to a value of the EPS strength
(and therefore of the electron mobility). Plain lines correspond to
the self-consistent calculations. Dashed lines correspond to the calculated
DOS using eq . The
conduction band edge has been shifted in energy to 0 eV.
Figure 4
Density of states in the considered device as a function
of the
electron energy. Each color corresponds to a value of the optical
phonon energy. Plain lines correspond to the self-consistent calculations.
Dashed lines correspond to the calculated DOS using eq . The conduction band edge have
been shifted in energy to 0 eV.
Density
of states in the considered device as a function of the
electron energy. Each color corresponds to a value of the EPS strength
(and therefore of the electron mobility). Plain lines correspond to
the self-consistent calculations. Dashed lines correspond to the calculated
DOS using eq . The
conduction band edge has been shifted in energy to 0 eV.Density of states in the considered device as a function
of the
electron energy. Each color corresponds to a value of the optical
phonon energy. Plain lines correspond to the self-consistent calculations.
Dashed lines correspond to the calculated DOS using eq . The conduction band edge have
been shifted in energy to 0 eV.As seen before, DOS tends to penetrate deeper in
the band gap as
the scattering with optical phonons becomes more important. Another
interesting and unexpected result shown Figure is that the polar optical phonon energy
also plays an important role. The higher ithe phonon energy, the deeper
the DOS expands below the band edge. This result confirms the fact
that the DOS penetration in the gap is due to the multiphonon process.
While with a smaller Epop, the impact
of scattering is expected to be larger, because the number of phonons
increases, we show that the consequence on the DOS is smaller. From
the Urbach tail point of view, the low value of Epop in perovskites appears to be finally an advantage.
Urbach Energy
Using our results, we can now estimate
the Urbach energies by calculating the exponential decay of the absorption
coefficient α. Starting from the calculation of the photogenerated
current Isc (eq ), we obtain α expressed as a function
of the photon energy hνwhere L is the thickness
of the device; ve(hν)
is the group velocity of electrons generated by a photon of energy hν; and a and dhν are the numerical steps in position and energy, respectively.
Applying this procedure to GaAs,[19] we find
the Urbach energy of 8 meV, which is in good agreement with the experimental
value of 7.5 meV.[18] For the perovskite, Figure shows α for
each CEPS as a function of the photon
energy. It confirms the Urbach-like exponential decrease of the DOS
inside the band gap, which is all the more severe as EPS is large.
The corresponding Urbach energies are reported in Table . Despite the very low Epop, the calculated EU are not negligible, up to 9.5 meV, compared to the experimental
value of 15 meV. Yet, this value is generally explained by crystalline
disorder; our result shows that EPS also impacts the Urbach behavior.
However, contrary to the disorder, scattering with phonons is intrinsic
and, therefore, unavoidable.
Figure 5
Absorption coefficient α (solid lines)
versus photon energy
calculated for each value of the electron–phonon scattering
strength. Colored dashed lines correspond to the slopes used to calculate
Urbach energies. Black dashed line correspond to the band gap of CH3NH3PbI3 at 1.55 eV.
Absorption coefficient α (solid lines)
versus photon energy
calculated for each value of the electron–phonon scattering
strength. Colored dashed lines correspond to the slopes used to calculate
Urbach energies. Black dashed line correspond to the band gap of CH3NH3PbI3 at 1.55 eV.
Emission and Voc
Figure shows the calculated
emission spectra for each value of the scattering strength, calculated
with eq . We report
both a red shift and an increase of the intensity with the EPS strength.
This is another consequence of the DOS penetration. Electrons in the
conduction band can occupy the states below the bottom of conduction
band to recombine with holes above the top of valence band, emitting
photons of lower energies than the material band gap. In an efficient
photovoltaic device, we want to absorb as many photons as possible
while preventing electron–hole recombinations as much as possible.
To quantify the impact of scattering on the open-circuit voltage,
we calculate the ratio of absorption to emission (needed in eq ). Figure a shows the calculation of this ratio as
a function of CEPS. The ratio decreases
for higher values of phonon coupling. The penetration of the DOS due
to EPS therefore favors the recombination and thus the photon emission.
This confirms that the degradation of the sharpness of the absorption
coefficient due to Urbach tail tends to deteriorate the quality of
the material for photovoltaic applications.[23] Decrease of the absorption-to-emission ratio with the EPS can be
linked to the Urbach behavior. Corresponding values of δoc are also shown in Figure b. We then confirm
that oc decreases with
the EPS up to 41 mV for the highest CEPS, therefore for the lowest mobility. Fortunately, this degradation
is limited by the low value of Epop in
the studied perovskite material. However, this value is not negligible
and such a behavior is intrinsic and generalizable to all semiconductors.
Figure 6
Emission
spectra calculated for each value of the electron–phonon
scattering strength.
Figure 7
Top: Isc/I0 ratio for each value of CEPS considered.
Bottom: oc losses due
to electron–phonon scattering for each value of CEPS considered.
Emission
spectra calculated for each value of the electron–phonon
scattering strength.Top: Isc/I0 ratio for each value of CEPS considered.
Bottom: oc losses due
to electron–phonon scattering for each value of CEPS considered.
Conclusions
In this paper, we studied the impact of
the electron–phonon
scattering on optical and electrical properties of an hybrid perovskite.
Our numerical calculations showed that electron–phonon scattering
induces a penetration of the density of states in the band gap, which
increases with the scattering strength. Such modifications induce
an energy shift of the emission spectra, since photogenerated electrons
and holes can recombine at lower energies than the material band gap.
We have also investigated the Urbach effect induced by electron–phonon
scattering. We obtained values of Urbach energies from 5.25 to 9.5
eV, which means that phonon scattering contributes significantly to
the exponential decrease of absorption below the band gap. The analytical
description of the density of states shows that the Urbach tail is
due to the multiphonon process, implying an unexpected dependence
on the phonon energy. The low value of this energy in the considered
material appears to be an advantage. Finally, we calculated the open-circuit
voltage Voc for each scattering strength
compared to the ballistic counterpart. We obtained Voc losses up to 41 mV due to the electron–phonon
scattering. To tweak the results concerning the impact of electron–lattice
interactions on the optical and electrical properties, a more precise
inclusion of CH3NH3 rotations should be led.
Moreover, the study of the influence of temperature can also provide
relevant information about the carrier-transport properties in such
materials.
Authors: Tom J Savenije; Carlito S Ponseca; Lucas Kunneman; Mohamed Abdellah; Kaibo Zheng; Yuxi Tian; Qiushi Zhu; Sophie E Canton; Ivan G Scheblykin; Tonu Pullerits; Arkady Yartsev; Villy Sundström Journal: J Phys Chem Lett Date: 2014-06-12 Impact factor: 6.475
Authors: Dong Shi; Valerio Adinolfi; Riccardo Comin; Mingjian Yuan; Erkki Alarousu; Andrei Buin; Yin Chen; Sjoerd Hoogland; Alexander Rothenberger; Khabiboulakh Katsiev; Yaroslav Losovyj; Xin Zhang; Peter A Dowben; Omar F Mohammed; Edward H Sargent; Osman M Bakr Journal: Science Date: 2015-01-30 Impact factor: 47.728
Authors: Jan Pospisil; Oldrich Zmeskal; Stanislav Nespurek; Jozef Krajcovic; Martin Weiter; Alexander Kovalenko Journal: Sci Rep Date: 2019-03-04 Impact factor: 4.379