Article

Cite This: J. Am. Chem. Soc. 2017, 139, 18262−18270 pubs.acs.org/JACS

Free Carrier Emergence and Onset of Electron−Phonon Coupling in

Methylammonium Lead Halide Perovskite Films
Tufan Ghosh, Sigalit Aharon, Lioz Etgar, and Sanford Ruhman*
Institute of Chemistry, The Hebrew University of Jerusalem, Jerusalem, Israel 91904
*
S Supporting Information

ABSTRACT: Sub-10 fs resolution pump−probe experiments

on methylammonium lead halide perovskite films are described.
Initial response to photoexcitation is assigned to localized hot
excitons which dissociate to free carriers. This is attested to by
band integrals of the pump−probe spectra where photoinduced
bleaching rises abruptly 20 fs after photoexcitation. Later stages
of spectral evolution are consistent with hot carrier cooling,
during which state filling induced bleaching of interband and
exciton transitions curiously more than doubles. Electron
coupling to optical phonons is observed as periodic spectral
modulations in the pump−probe data of both films. Fourier
analysis identifies active phonons at ∼100 and 300 wavenumbers pertaining to the lead-halide framework and organic cation
motions, respectively. Coupling strengths estimated from the depth of these modulations are in the weak coupling limit, in
agreement with values extracted from temperature dependent emission line shape analysis. These findings support free carriers in
these materials existing as large polarons. Accordingly, these modes are probably not dictating the moderate carrier mobility in
this material.

■ INTRODUCTION
The impressive power conversion efficiency (∼20%) obtained
renormalization by hot carriers. These assignments are based
on the dependence of these features on pump fluence and
in solar cells based on three-dimensional organic lead halide wavelength. Finally, several interpretations of the delayed broad
perovskites (OLHP), along with their ease of preparation, have PIA higher in the interband continuum have been suggested.
generated an immense effort to understand the underlying Some have assigned this to the same band gap renormalization
photophysics. Results show that crucial factors behind their mentioned above, while others trace it to additional intraband
high power conversion efficiency are large optical absorption excitations. All of these need to be re-examined in light of a
cross sections, substantial electron−hole diffusion lengths and recent study which assign this band to pump induced variations
slow electron−hole recombination rates.1−10 Given OLHP’s in reflectivity of the sample and not related to bulk absorption
hybrid nature, assessing the roles of exciton vs free carrier changes.17
transport has been an issue of central importance. While views Most pump−probe experiments on OLHPs were conducted
on this vary, most studies highlight the essential role of free with time resolution incapable of recording initial carrier
electrons and holes in the context of photovoltaic perform- emergence and thermalization. This includes a previous study
ance.11,12 Under solar irradiation, most photons absorbed in from this lab which showed that much of the BE PIB appears
methylammonium lead iodide perovskite (MAPbI3) possess before substantial cooling of the nascent hot carriers and this
∼0.4 eV of excess energy above the 1.65 eV band gap (BG). was assigned to effects of hot carrier screening on the state
Accordingly characterizing the dynamics of free carrier density near the BE.20 The spectral tailing of these screening
generation and relaxation is of central importance, requiring effects was employed to estimate an exciton binding energy of
ultrafast spectroscopic investigation. 20−30 meV for both MAPbI3 and MAPbBr3 layers. But even
A number of dynamic spectral features are common to the improved time resolution employed there did not uncover a
femtosecond pump−probe studies on OLHP films: (1) a rapid signature of exciton dissociation into free carriers. A recent
buildup of intense photoinduced bleach (PIB) centered at the study has addressed these questions using extreme time
optical band-edge, which broadens with increasing pump resolution electronic 2D spectroscopy21 from which carrier
fluence, (2) the appearance of a short-lived photoinduced thermalization is concluded to be the fastest process observed,
absorption (PIA) band to the “red” of the optical band edge with a duration which is highly dependent on the excitation
(BE), which decays over the course of hot carrier cooling, and photon energy. However, the intensity distribution of the pulse
(3) a slowly rising low intensity broad PIA near ∼650 nm.13−20
Scenario (1) is often assigned to low energy state filling, and Received: September 6, 2017
scenario (2) to shifting of exciton states due to band gap Published: November 26, 2017

© 2017 American Chemical Society 18262 DOI: 10.1021/jacs.7b09508

J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

Figure 1. (a) Ground state absorption spectrum of MAPbI3 (black line) and NOPA pump spectrum (gray). (b) Same as (a) for MAPbBr3..

Figure 2. (a, c, e) Transient transmission, finite difference, and species associated spectra of MAPbI3 films after excitation to a carrier density of 1 ×
1018 cm−3. (b, d, f) Same for a carrier density of 5 × 1018 cm−3. Associated decay constants for (e) and (f) are presented in Table 1.

spectrum did not allow coverage of the band edge features, Here we report a sub-10 fs pump−probe study of
which are both highly structured and exhibit much of the early microstructured MAPbI3 and MAPbBr3 films. Data analysis
evolution of interest. In particular two of the three features proves that sample bleaching rises abruptly with an ∼20 fs delay
common to ultrafast studies of OLHPs involve this transition following photoexcitation which is assigned to dissociation of
explicitly. Thus, despite the extreme time resolution, the results localized and correlated hot electron hole pairs. The overall
of this experiment do not provide a full spectral mapping which bleaching of exciton and interband absorption later rises
can conclusively analyze such high time resolution experiments exponentially by a factor of 2.5 with a 250 fs lifetime, tentatively
in terms of the carrier dynamics they reflect. assigned to increased state filling of the BE exciton during
There is another material response which is expected to carrier cooling. Spectral modulations in pump−probe signal
affect spectral evolution of the exciton transition at early times due to impulsively excited LO phonons are recorded and used
to estimate the strength of electron-optic phonon coupling in
which has strangely not been observed in pump−probe
both materials. The resulting coupling strengths are well within
experiments to date, and that is the excitation and relaxation
the weak regime and compatible with conduction electrons
of optical phonons. The importance of optic phonon coupling existing as large polarons.

■
to the conduction of free carriers in these materials has been
extensively discussed in recent publications.22−31 While the
MATERIALS AND METHODS
magnitude of electron phonon coupling in OHLPs has been
tested by following the temperature dependence of emission Thin films of methylammonium lead-halide perovskites were prepared
according to a previously described method.34 In brief, first spin
line shapes, few time domain measurements have reported coating of 1 M solutions of PbX2 (X: I and Br) in dimethylformamide
impulsive phonon activity like that observed here, and those (Sigma-Aldrich) was carried out on 0.5 mm thick glass substrates. The
few have not converged in terms of the active phonon spin velocity was 6500 rpm and the spin-coating duration was 5 s.
frequencies or their damping dynamics.32,33 After the spin-coating, the substrates were annealed at 80 °C for 10

18263 DOI: 10.1021/jacs.7b09508

J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

min. In the second step, the spin coated PbX2 films were dipped into a hν τ1 τ2 τ3
solution of 63 mM CH3NH3X (X: I and Br) in isopropanol. Following to the following kinetic scheme: →A → B → C → D, leading
the dipping step, the samples were annealed at 80 °C for another 30 to the evolution associated spectra shown in the bottom row of
min. The samples were characterized by steady state absorption and Figure 2. Proper curve fitting required a minimum of five
photoluminescence spectroscopy. The visible absorption spectra were components; an instantaneous coherent coupling contribution
measured using a Jasco V-670 spectrophotometer and photo- (CC in bottom row of Figure 2, not included in the kinetic
luminescence measurements were carried using a fluorescence scheme above), followed by a sequence of four intermediates,
spectrometer (Edinburgh Instruments FLS920) equipped with a film all convoluted with a 15 fs Gaussian instrument response
holder. The pump−probe experiments were performed with a home-
built amplified Ti-sapphire laser producing 1 mJ 30 fs pulses centered function. This kinetic model provided near perfect fits to the
at 790 nm, at repetition rate of 1 kHz. The amplified output was split low noise data (Figure S2 in the Supporting Information), and
in two parts for the generation of the pump and probe, respectively. resulted in the time constants presented in Table 1.
Sub-10 fs pump pulses were generated in a noncollinear optical
parametric amplifier (NOPA) centered either at ∼530 nm (for Table 1. Decay Time Constants for Transitions between the
MAPbI3 experiments) or ∼490 nm (for MAPbBr3 experiments) as Species Associated Spectra Presented in Figure 2e and f
shown in Figure 1. The pump was compressed to within 15% of the
transform limit using a prism pair (BK7), followed by an aluminum carrier density, cm−3 τ1, fs τ2, fs τ3, fs
coated deformable mirror. Supercontinuum probe pulses were 1 × 1018 135 ± 15 180 ± 10 270 ± 15
generated by focusing the 1400 nm output of an optical parametric 2.4 × 1018 120 ± 10 220 ± 10 285 ± 15
amplifier (TOPAS 800, Light Conversion) on 2 mm of BaF2 (in the
5 × 1018 90 ± 10 285 ± 10 320 ± 15
case of MAPbI3 experiments), or by focusing a 800 nm beam on an ∼2
mm sapphire disk (for MAPbBr3 experiments). Thereafter, the pump
and the probe were directed toward the sample using reflective optics.
The carrier density dependent experiments were performed by Inspection of the data closely reveals faint intensity
changing the pump fluence. Figure 1 illustrates absorption spectra of modulations with periods of 100−200 fs. These are assigned
the perovskite samples, along with the corresponding pump spectra to coherent wave packets in optical phonons due to their
used for the photoexcitation. The carrier densities immediately after coupling to the electronic transition. To test the generality of
excitation (n0) were calculated for MAPbI3 as described in the this observation, similar experiments were conducted on
Supporting Information. MAPbBr3 thin films exciting just above the BE (Figure 3).

■ RESULTS
Pump−probe spectra of MAPbI3 upon NOPA photoexcitation
This also served to test if, as observed in CdSe nanocrystals,
coherent phonon modulations become stronger as excitation
photon energy approaches the fundamental gap.35 The
are shown in Figure 2 for two carrier densities (see Figure S1 in modulations were extracted by subtracting a background from
the Supporting Information for an additional intermediate the decay components at all probe wavelengths (Figure 4c and
density) along with sample and excitation pulse spectra. The
general trends described in the introduction are apparent in
both. The immediate response of photoexcitation is a red-shift
of the absorption with partial bleach at the BE. Over the next
∼1−2 ps, this red-shifted absorption diminishes and at the
same time a net bleach builds up at the BE. To isolate various
stages of spectral evolution, finite difference spectra are
calculated according to eq 1 and presented in the second row
of Figure 2.
⎛ Δt ⎞⎟
ΔΔOD(λ , t , Δt ) ≡ ΔOD⎜λ , t +
⎝ 2 ⎠
⎛ Δt ⎞⎟
− ΔOD⎜λ , t −
⎝ 2 ⎠ (1)
Noteworthy aspects of the finite difference spectra are (a) upon
absorption at both intensities a zero integral peak shift rises
sharply at the band edge, and surrounding the second
absorption band at 480 nm. (b) An ultrafast buildup of
bleaching throughout the interband continuum, peaking at the
BE, follows. This bleach red-shifts and narrows over ∼200 fs,
erasing the feature described in (a), changes which are assigned
to carrier cooling. (c) Concomitant transmission changes
surrounding the 480 nm peak differ in that no evolution in this
feature is apparent over the carrier cooling stage. Furthermore,
its intensity and shape are nonlinearly dependent on the pump
fluence. (d) The shifting described in (b) reveals an underlying
reduction in transmission over the range from 550 to 650 nm. Figure 3. (a) Transient difference and (b) finite difference
(e) All spectral features broaden as the excitation intensity is transmission spectra of MAPbBr3 films after photoexcitation with
increased. 490 nm NOPA pulses. (c, d) Temporal changes at the BE bleach and
In order to characterize the spectral evolution in the 480− below BE PIA for MAPbBr3 and (e, f) temporal changes at the BE
900 nm range, global kinetic analysis was conducted according bleach and below BE PIA for MAPbI3 films.

18264 DOI: 10.1021/jacs.7b09508

J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

Figure 4. (a, b) Two dimensional map of Fourier power for MAPbBr3 and MAPbI3 films. The abscissa depicts dispersed probe wavelength, while the
ordinate indicates the Fourier frequency. The Fourier power is color coded according to the rulers to the right of each panel. Black lines show the
power as a function of probe wavelength for the two prominent modulation peaks. (c, d) Residual modulations showing a π-phase shift when probed
at two different wavelengths for high frequency modes in MAPbBr3 and in MAPbI3.

d). Fourier analysis was performed to obtain the frequencies of The second facet is the discovery of a brief initial state of the
the associated phonons. The left panel in Figure 4 presents the excited sample, less than 20 fs in duration, which precedes any
results in color coded maps. Overlaid are cuts which present the phase of absorption bleaching despite excitation high above the
power as a function of the probe frequency for both prominent ∼1.65 eV band edge. Assignment of this phase to the break up
optic phonon peaks observed, at 90 ± 10 and 340 ± 15 cm−1 of localized carrier pairs is based on its briefness and on the
for MAPbBr3, and at 110 ± 10 and 235 ± 15 cm−1 for MAPbI3. sudden buildup of sample bleach which followes it. This
The low frequency band is tentatively assigned to displace- observation is novel not only to experiments on OLHPs, but to
ments in the lead-halide scaffolding, while the higher bulk semiconductors in general, making the interpretation of its
frequencies to motions in the methylammonium ions.36−38 In spectral fingerprint a challenge as shown bellow.
both cases, the Fourier mapping shows a doubled peak Emergence of Hot Carriers in MAPbI3. One objective
structure along the probe wavelength axis with humps lying here was to reveal previously unrecorded early stages of carrier
to either side of the exciton peaks in both materials, exhibiting generation in OLHP films. In order to achieve this, laser pulses
varying degrees of symmetry. shorter than 10 fs have been applied in these experiments. The

■
phase of spectral evolution unique to this study can be
DISCUSSION identified in two measures, species “A” in the kinetic analysis
(bottom row of Figure 2) or, equivalently, the black ΔΔOD
Two novel facets of OLHP photophysics have been uncovered spectra (middle row of Figure 2). Both spectra consist of sharp
here. One is the detection of new vibrational modes which peak shift signatures at the edges of both bands. In the finite
modulate the transient transmission spectra. From the difference spectra, a similar peak shift signature appears in the
modulation depth of the change in probe transmission, 480 nm band as well. According to the global kinetic scheme
coupling strengths to the exciton transition have been and its associated rates, this signature rises instantaneously and
estimated. This discovery did not depend on high time decays within ∼100 fs. However, despite the high quality of
resolution, but was facilitated by improved S/N in the fitting provided by the global analysis, similarity of decay times
pump−probe experiments. A significant characterization and associated with consecutive transitions, as well as deviations at
analysis of this finding stems from the recent interest in the very early delays, demonstrate that a kinetic scenario with
impact of lattice deformations on charge mobility and decay distinct species and exponential decays does not capture the
dynamics in these materials. Any new input to assess these underlying dynamics of this system. These results are presented
effects should prove valuable. nonetheless to comment on its limitations in light of the
18265 DOI: 10.1021/jacs.7b09508
J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

numerous application of this method in previous studies. Our

interpretation will therefore concentrate primarily on the finite
difference spectra and spectral band integrals to be described
below.
The newly revealed spectral snapshot should be interpreted
in context with the established chronology of hot carrier
dynamics in bulk semiconductors, starting with exciton
dephasing and sub-100 fs thermalization of holes and
conduction electrons whose initial energy is dictated by the
pump spectrum and carrier effective masses,39−42,10 followed by
picosecond cooling to the lattice temperature through phonon
scattering. Sharp peak shifting in ultrafast transient spectra is
often related to nonlinear coupling of pump and probe during
their temporal overlap, or until the pump induced transition Figure 5. Plot of band integrals for MAPbI3 from 550 to 900 nm
dipoles have dephased. We can exclude this assignment in the calculated according to eq 2. Exponential fit to the delayed rise is
present data since these shifts decay only on much longer time overlaid in red. Inset shows initial 60 fs alongside the rises of the below
scales. To demonstrate this, Figure S3 in the Supporting band edge PIA (A1) and BE bleach signals during the same period.
Information presents a series of finite difference spectra where
ΔOD (t = 15 fs) is subtracted from a series of later transient temporal overlap of pump and probe. This can be seen when
spectra extending up to 100 fs. These are depicted along with comparing the rise in ΔD with that of both halves of the BE
the 15 fs transient transmission spectrum. Clearly the ultrafast shift feature, included in the inset to Figure 5. In bulk
shift does not decay away over the first tens of femtoseconds as semiconductors carrier thermalization has been determined to
expected from electronic coherent coupling. This ultrafast take place on similar time scales.39−42 However, since the pump
spectral shifting seemingly precedes thermalization since spectrum overlaps significantly with the integration range of
spectral variations during carrier cooling should be gradual. In ΔD, state filling should be present instantly. Accordingly, the
contrast, shifts in difference spectra over the initial ∼100 fs of delayed ultrafast appearance of band bleaching is assigned to
delay shown by species “B” or by the red ΔΔOD curves, reveal the emergence of uncorrelated free carriers through exciton
a sudden shift to net bleaching. These are assigned to state dephasing and dissociation. This roughly equals twice the time
filling, consistent with the establishment of a hot Fermi−Dirac it takes nascent carriers to separate a distance equal to the
carrier distribution. average spacing between the photogenerated pairs at the
The sharp red-shift at the band edge can be assigned to a densities tested. This delay is also on par with the dephasing
shift in exciton and not interband transitions. Stokes shifting of times determined using 2D spectroscopy,21 suggesting exciton
a step-function-like interband absorption would induce dissociation is the main mechanism of decoherence high above
absorption below the edge without a concomitant induced the BG. This appearance time is much shorter than that
transparency above, contrary to the observed trend. Further- reported from broadband THz probing studies.43 Observing
more, an integral of the shift feature near 750 nm is centered at this phase hinges on the unique capability of the ultrashort
and of the same spectral width of the exciton band (Figure S4 pump pulses to generate localized e−h pairs even high above
in the Supporting Information) as isolated by modeling of the BG due to coherent excitation of a broad band of “k” states.
MAPbI3 linear absorption spectra in numerous publica- The source of delay in bleach appearance while electrons and
tions.12,18 The remarkably good match supports the above holes are correlated will require further investigation. It may be
assignment. due to the crossed polarization of pump and probe chosen to
In order to distinguish between exciton dissociation to carrier suppress scattering. However, previous photoselective pump−
thermalization, and demonstrate the delayed nature of the probe experiments on perovskites exhibited very limited effects
bleach buildup, difference dipole strength band integrals were at later stages of evolution.14
extracted from the data covering the range from 550 to 900 nm Understanding the spectral signature of the dissociating pairs
(Figure 5) according to eq 2: is difficult since very few equivalent studies have been
900 ΔOD(ν , t ) conducted in ordinary inorganic semiconductor solids. Much
ΔD(t ) ≡ ∫550 ν
dν
(2)
of the ultrafast study of carrier dynamics in bulk semi-
conductors from the 1980s and 1990s were conducted by
This measure can detect the onset and evolution of state filling pioneers of ultrafast spectroscopy before advent of the tools we
without reference to specific structure in the transient have at our disposal, limiting time resolution by a factor of 2−4,
spectrum. The lower cutoff at 550 nm provides selective and excluding the flexibility of tuning both of pump and of
detection of excitation effects on sample absorption across the probe. Nonetheless, the ground-breaking efforts to resolve the
fundamental gap without interference from higher transitions. earliest stages of free carrier generation by Hu et al.44 and Bar-
Remarkably, despite significant pump intensity dependence of Ad and Chemla45 concur with our assignments. Hu et al.
the TA data in Figures 2a,b, 3e,f, S1 and S5 (Supporting identify an instantaneous polarization of the electron hole pairs
Information) the time dependence of these band integrals is even before they have time to separate using THz probing, and
identical within error for all three pump fluences (Figure S6 in assign a characteristic time scale of ∼10 fs. This is in GaAs
the Supporting Information). It consists of an abrupt rise where the binding energy is minute and this stage should
followed by a gradual growth in the net bleach by a factor of 2.5 accordingly be shorter than in the OLHP materials. Chemla
which is well fit by an exponential of τ = 240 fs. and Bar-Ad applied an approach similar to our own but
Instead of rising instantly, the sudden onset in bleach is limitation of probe bandwidth and pulse durations did not
delayed by nearly 20 fs relative to time zero, defined as allow full coverage of the band edge exciton peak fully. It would
18266 DOI: 10.1021/jacs.7b09508
J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

seem however that a fast peak-shift is the leading material We note that a similar decay in biexciton shifting at the band
response there as well, similar to our own. edge is also observed in pump−probe experiments on
Aside from a firm assignment of the earliest spectral shifting semiconductor nanocrystals.52−54
to the exciton band, interpretation of its origin, and why it Comparison of transmission changes surrounding the
persists well beyond the stage which we assign to dissociation fundamental band edge with the 480 nm peak shows that the
will require identifying whether this feature is related to ratio of intensities of these signatures decreases markedly with
changes in absorption or in reflection. Possible mechanistic increasing pump fluence. The latter which is nearly
connection between the peak shift observed at early times to a unobservable at the lowest pump intensity becomes half as
similar feature observed in reflectivity changes in electro- strong at the higher fluence as the shift band around 750 nm.
absorption experiments46−48 will have to await such an Interpreting this requires a separation of pump induced changes
identification as well. It is possible that the enhanced in transmission and reflection with the current time resolution
polarizability of the excited pairs causes an effective rise in and spectral coverage. Furthermore, the 480 nm band is
sample dielectric constant providing attractive biexciton curiously unaffected by carrier cooling at either pump intensity
interaction lowering the resonance frequency of further exciton in contrast to the significant transmission changes near 750 nm.
absorption. The slow decay of this feature may indicate that This points to asymmetry in the dynamics of carrier cooling
such stabilization can also be affected by carriers which have between the conduction and valence bands, only one of which
lost correlation as well. In any case, it makes perfect sense that is suggested to be mutual to both transitions.55 This is
state filling effects kick in only once the nascent pair coherences unexpected since effective masses are similar in conduction and
dephase and populations are established. valence bands.
The next stage of spectral evolution complete in roughly 100 The gradual rise exhibited by ΔD over hundreds of
fs is assigned to establishment of a thermal distribution of hot femtoseconds requires further comment. Modeling of band
carriers. This delay is longer than reported for some inorganic filling effects is usually performed assuming independence of
semiconductors. The only experiment conducted with similar the transition dipole on k,⃗ producing interband absorption
time resolution on this material employing 2D spectroscopy21 which is linear in the joint density of states.18 If this approach
extracted pump wavelength dependent thermalization times in were rigorous and absorption bleaching was due exclusively to
the 10−50 fs range, in fair agreement with observations here. blocking of interband transitions, the integrated bleach would
Cooling of Hot Carriers. Later stages of spectral evolution be constant. This strong deviation may reflect a combination of
consist of a gradual shift to the blue and a narrowing of the three possible factors. On the one hand, the assumption of
bleach from shorter to longer probe wavelengths. This shift independence of mVC on either may be incorrect. Another
uncovers an underlying net reduction in transmission in the possibility is that the enhanced apparent bleach is due to the
range from 510 to 730 nm, assignment of which is the subject buildup of stimulated emission which appears once carriers in
of debate.17,18 Deschler and co-workers, who separately both bands have relaxed to the band edge. The third involves
measured transient transmission and reflection, have attributed the effects of an increasing population and blocking of
this to pump induced changes in reflectivity.17 Yang et al. who absorption to the n = 1 exciton band which accompanies
also measured reflectivity effects, concluded instead that BG carrier cooling.51 In any case, this observation questions
renormalization is responsible to this broad PIA.18 A recent interpretations of pump−probe data in terms of bleach
report by Anand et al. assign the broad induced absorption to conserving models for extracting the time dependence of
the presence of additional valence and/or conduction bands.49 carrier temperatures.
Since reflectivity was not addressed here, we cannot contribute Electron−Phonon Coupling in MAPbBr3 and MAPbI3
to this debate. The trend of continuous red-shifting and Films. Nascent free carriers also couple to optical phonons via
narrowing of a bleach toward the BE is in line with cooling the Fröhlich mechanism which has recently been investigated
induced concentration of the photoinduced carriers at the by many groups.22−27,32,33,56−58 All subsequent dynamical
bottom of the conduction and valence bands.15−18,20 processes, including mobility and recombination, are influenced
One noteworthy aspect concerns the near linearity of the by the strength of this interaction, making its quantification
amplitude of post cooling transient spectra with pump fluence essential. In this section, we will focus on the signatures of
as demonstrated in Figure S7 in the Supporting Information. coherent underdamped optical phonons on our measurements
This proves that within the range of carrier densities and how it can serve to estimate the coupling strength to these
experimented with here, population of the exciton states modes. In the pump−probe data, we observed spectral
remains limited, without significant Burstein−Moss shifting. modulations surrounding the band edge both in MAPbBr3
This is surprising since densities here approach one excitation and MAPbI3 due to electron−phonon coupling. The periodic
per Bohr volume. Even with significant thermal dissociation of modulations in the residuals were obtained by subtracting a
excitons at room temperature, transient spectra should exhibit background from the transient kinetics (Figure 4c and d).
some saturation of the band edge bleach. This, however, Fourier analysis of the residuals identifies frequencies of the
depends on the exciton binding energy, estimates of which vary phonons and the intensity of modulation at various probe
widely.46,50,51 wavelengths.
Another aspect is the disappearance of the exciton state peak For MAPbBr3, the phonon frequencies are 90 ± 10 and 340
shift as carrier cooling advances at all three excitation densities. ± 15 cm−1 (Figure 4a). The low frequency band has been
This feature is present for at least 100 fs after rising abruptly assigned to optical phonons involving the lead−halide
and is assigned to screening effects on the excitonic peak. Why scaffolding, and the higher frequencies to motions in the
this shift decays during cooling cannot be explained by methylammonium ion.36−38 The two clearly observed phonon
complete state filling of the exciton band since even at the frequencies in MAPbI3 are 110 ± 10 and 235 ± 15 cm−1
highest pump fluence the combined losses due to changes in (Figure 4b). In both cases faint additional peaks are observed
absorption and reflection are a small fraction of the sample OD. between these prominent features, whose identity is the subject
18267 DOI: 10.1021/jacs.7b09508
J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

of ongoing study. Furthermore, exciting MAPbI3 at the band Table 2. Electron−Phonon Coupling Parameters
gap showed similar modulation amplitudes as excitation high Determined for MAPbI3 and MAPbBr3
above it, as presented in Figure S8 in the Supporting
ℏωLO (meV) S Δ λ (meV)
Information, contrary to observations in semiconductor
MAPbI3 29.13 0.04 0.3 1.1
nanocrystals.35
13.63 0.02 0.2 0.3
The identity of the high frequency mode previously assigned
MAPbBr3 42.15 0.08 0.4 3.5
to torsions of the MA cation is subject to some debate. The
11.4 0.09 0.4 1.1
large shift of this mode for different halide perovskite has been
suggested to reflect strong interaction between the PbX3− and
CH3NH3+ resulting in a restricted rotation of the methyl
around its C−N axis.59 However, a recent study has suggested These results join those of a growing number of studies
that this mode is no longer valid for the MA cation torsion addressing OLHP electron phonon coupling, whose findings
since it show up the Raman spectra of both MAPbBr3 and have yet to converge in terms of frequencies, coupling
CsPbBr3 perovskites.60 As for the low frequency modulations strengths, and mode dynamics. The interest in this aspect of
observed in the two samples, that observed for the bromide falls OLHP photophysics stems from the realization that despite the
below that in the iodide. There is nothing in our experiments predominant involvement of the inorganic framework in
which can explain this puzzling result. Either this mode involves defining their electronic structure, carrier mobilities fall midway
negligible motion of the halide, or is fundamentally different in between those observed in conventional inorganic semi-
the two materials. Answering this will require further conductors and in organic conductors. Other underdamped
investigation. modes likewise revealed through impulsive spectroscopy vary in
Impulsive excitation of molecular chromophores induced frequency from tens to several hundred wavenumbers, while
coherent vibrations in the excited state by simultaneous frequency domain spectroscopies highlight modes in the 10−50
population of numerous vibronic levels in the form of a cm−1 range which are predominantly overdamped at room
localized wave packet, and in the ground state via impulsive temperature.57 Full comprehension of the role played by all
resonance Raman scattering. As shown in Figure 4a, the these vibrations in any particular photophysical attribute of
modulation amplitudes peak in two lobes on either side of the OLHP films will require consolidation of these find-
bromide exciton peak, with 1π phase shift on either side (Figure ings.23,25,32,62 Toward establishing such consolidation it is of
4c and d). This indicates that the effect of the vibration is utmost importance that all empirical findings concerning
primarily to shift the central wavelength of the exciton frequencies and coupling strengths of active modes be reported
transition. In molecules, excited state and ground state in the literature. Nonetheless, assuming the moderate charge
populations are distinct, a separation which is more difficult mobilities in OLHPs has been correctly assigned to
in solids where our understanding is based on delocalized intermediate electron phonon coupling, the modes reported
here would appear not to be responsible for this effect.

■
excitations. Rather than untangling this distinction here, a
heuristic analysis similar to that described for quantum dots by
Kambhampati and co-workers is applied.35 CONCLUSION
A displaced harmonic oscillator model for the band edge High signal-to-noise pump−probe measurements with sub-10
exciton transition is envisioned, with the depth of modulation fs pulses were performed to uncover early stages of carrier
obtained by comparing the modulation amplitudes to that of dynamics and electron-optic phonon coupling in lead halide
the bleach itself. Electron−phonon coupling strengths are perovskite films. Immediately after photoexcitation, localized
represented by a set of parameters [Δ, S, λ]. Δ is a hot excitons are generated which exhibit a sharp red-shift of the
dimensionless displacement of the normal coordinate, S is the excitonic peak in the TA spectra. These localized hot excitons
Huang−Rhys parameter, and λ is the reorganization energy. S is then dissociate to free carriers with a time scale ∼20 fs followed
related to Δ as S = Δ2/2, and the reorganization energy λ to S by thermalization which is complete within 100 fs as evidenced
as λ = ℏωLOS, where ωLO is the frequency of the optical in band integrals and global kinetic analysis. Later, hot carriers
phonon. λ (ℏΔω) was calculated from our data using cool to lattice temperature through scattering with optic
phonons within 0.5 ps. The charge carriers also couple to the
A OSC = ( ddODω )Δω, where OD is the optical density of the optic phonons which result in spectral modulations surround-
sample and AOSC is the amplitude of the oscillations in the same ing the band edge in both the materials. Fourier analysis of
units. d OD/dω was obtained from excitonic absorption spectra these modulations shows active phonons with frequency ∼ 100
from the literature normalized to 20 mOD for MAPbI3 and 30 cm−1 related to stretching modes of Pb-halide moieties and in
mOD for MAPbBr3 (amplitude of the band edge bleach in our the 250−350 cm−1 range previously assigned to torsional
motion of the methylammonium cation. The amplitudes of
data).18,61 d OD was obtained by numerical differentiation. AOSC these modulations allowed estimates for the underlying
dω
of the high frequency modes (∼340 and ∼235 cm−1) was electron phonon coupling strengths showing all to be well
obtained by fitting the residuals to a damped sine function within the weak regime and compatible for large polaron
which was then extrapolated to t = 0. Amplitudes of the low formation in these materials, and thus probably not responsible
frequency modes were estimated directly from the scans. The to the moderate carrier mobilities in these materials.
resulting coupling strengths, displacements and reorganization
energies for both materials and modes are provided in the
Table 2. Since S < 1 in both materials, weak electron−phonon
■
*
ASSOCIATED CONTENT
S Supporting Information
coupling of the detected underdamped modes predicting The Supporting Information is available free of charge on the
formation of large polarons in both materials. ACS Publications website at DOI: 10.1021/jacs.7b09508.
18268 DOI: 10.1021/jacs.7b09508
J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

Estimation of carrier density, fitting of TA data using (17) Price, M. B.; Butkus, J.; Jellicoe, T. C.; Sadhanala, A.; Briane, A.;
Global kinetic analysis model and other spectroscopic Halpert, J. E.; Broch, K.; Hodgkiss, J. M.; Friend, R. H.; Deschler, F.
Nat. Commun. 2015, 6, 8420.
data (PDF) (18) Yang, Y.; Ostrowski, D. P.; France, R. M.; Zhu, K.; van de

■
Lagemaat, J.; Luther, J. M.; Beard, M. C. Nat. Photonics 2016, 10, 53−
AUTHOR INFORMATION 59.
(19) Chen, K.; Barker, A. J.; Morgan, F. L. C.; Halpert, J. E.;
Corresponding Author Hodgkiss, J. M. J. Phys. Chem. Lett. 2015, 6, 153−158.
*[email protected] (20) Sharma, V.; Aharon, S.; Gdor, I.; Yang, C.; Etgar, L.; Ruhman, S.
J. Mater. Chem. A 2016, 4, 3546−3553.
ORCID
(21) Richter, J. M.; Branchi, F.; Valduga de Almeida Camargo, F.;
Tufan Ghosh: 0000-0002-4898-2117 Zhao, B.; Friend, R. H.; Cerullo, G.; Deschler, F. Nat. Commun. 2017,
Lioz Etgar: 0000-0001-6158-8520 8, 376.
Sanford Ruhman: 0000-0003-0575-1367 (22) Zhu, H.; Miyata, K.; Fu, Y.; Wang, J.; Joshi, P. P.; Niesner, D.;
Williams, K. W.; Jin, S.; Zhu, X.-Y. Science 2016, 353, 1409−1413.
Notes (23) Wright, A. D.; Verdi, C.; Milot, R. L.; Eperon, G. E.; Pérez-
The authors declare no competing financial interest. Osorio, M. A.; Snaith, H. J.; Giustino, F.; Johnston, M. B.; Herz, L. M.

■ ACKNOWLEDGMENTS
S.R. holds the Lester Aronberg Chair in Chemistry. S.R.
Nat. Commun. 2016, 7, 11755.
(24) Straus, D. B.; Hurtado Parra, S.; Iotov, N.; Gebhardt, J.; Rappe,
A. M.; Subotnik, J. E.; Kikkawa, J. M.; Kagan, C. R. J. Am. Chem. Soc.
2016, 138, 13798−13801.
acknowledges support from the US-Israel Binational Science (25) Mante, P.-A.; Stoumpos, C. C.; Kanatzidis, M. G.; Yartsev, A.
foundation and the Israel Science Foundation. S.R. thanks Prof. Nat. Commun. 2017, 8, 14398.
M. Bonn, S. Koch, L. Kronik, R. Baer and doctors O. Yaffe, J. (26) Sendner, M.; Nayak, P. K.; Egger, D. A.; Beck, S.; Müller, C.;
Ruhman, and B. Skinner for enlightening discussions. T.G. Epding, B.; Kowalsky, W.; Kronik, L.; Snaith, H. J.; Pucci, A.;
thanks the Lady Davis Fellowship Trust and the Raymond and Lovrinčić, R. Mater. Horiz. 2016, 3, 613−620.
Janine Bollag Post-Doctoral Fellowship Fund for a fellowship. (27) Monahan, D. M.; Guo, L.; Lin, J.; Dou, L.; Yang, P.; Fleming, G.
S.R. and T.G. thank Dr. Oleg Liubashevski for technical R. J. Phys. Chem. Lett. 2017, 8, 3211−3215.
assistances. L.E. acknowledges the China-Israel grant under the (28) Ivanovska, T.; Dionigi, C.; Mosconi, E.; De Angelis, F.; Liscio,
F.; Morandi, V.; Ruani, G. J. Phys. Chem. Lett. 2017, 8, 3081.
Israel Ministry of Science.

■
(29) Wolf, C.; Cho, H.; Kim, Y.-H.; Lee, T.-W. ChemSusChem 2017,
10, 3705.
REFERENCES (30) Karakus, M.; Jensen, S. A.; D’Angelo, F.; Turchinovich, D.;
(1) Lee, M. M.; Teuscher, J.; Miyasaka, T.; Murakami, T. N.; Snaith, Bonn, M.; Cánovas, E. J. Phys. Chem. Lett. 2015, 6, 4991.
H. J. Science 2012, 338, 643−647. (31) Bonn, M.; Miyata, K.; Hendry, E.; Zhu, X.-Y. ACS Energy Lett.
(2) Grätzel, M. Nat. Mater. 2014, 13, 838−842. 2017, 2, 2555.
(3) Green, M. A.; Ho-Baillie, A.; Snaith, H. J. Nat. Photonics 2014, 8, (32) Batignani, G.; Fumero, G.; Kandada, A. R. S.; Cerullo, G.;
506−514. Gandini, M.; Ferrante, C.; Petrozza, A.; Scopigno, T.
(4) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. J. Am. Chem. Soc. arXiv:1705.08687, 2017.
2009, 131, 6050−6051. (33) Miyata, K.; Meggiolaro, D.; Trinh, M. T.; Joshi, P. P.; Mosconi,
(5) Jeon, N. J.; Noh, J. H.; Kim, Y. C.; Yang, W. S.; Ryu, S.; Seok, S. I. E.; Jones, S. C.; De Angelis, F.; Zhu, X.-Y. Sci. Adv. 2017, 3, e1701217.
Nat. Mater. 2014, 13, 897−903. (34) Aharon, S.; Cohen, B. E.; Etgar, L. J. Phys. Chem. C 2014, 118,
(6) Kim, H.-S.; Lee, C.-R.; Im, J.-H.; Lee, K.-B.; Moehl, T.; 17160−17165.
Marchioro, A.; Moon, S.-J.; Humphry-Baker, R.; Yum, J.-H.; Moser, J. (35) Sagar, D. M.; Cooney, R. R.; Sewall, S. L.; Dias, E. A.; Barsan, M.
E.; Grätzel, M.; Park, N. − G. Sci. Rep. 2012, 2, 591. M.; Butler, I. S.; Kambhampati, P. Phys. Rev. B: Condens. Matter Mater.
(7) Stranks, S. D.; Eperon, G. E.; Grancini, G.; Menelaou, C.; Phys. 2008, 77, 235321.
Alcocer, M. J. P.; Leijtens, T.; Herz, L. M.; Petrozza, A.; Snaith, H. J. (36) Quarti, C.; Grancini, G.; Mosconi, E.; Bruno, P.; Ball, J. M.; Lee,
Science 2013, 342, 341−344. M. M.; Snaith, H. J.; Petrozza, A.; Angelis, F. D. J. Phys. Chem. Lett.
(8) Burschka, J.; Pellet, N.; Moon, S.-J.; Humphry-Baker, R.; Gao, P.; 2014, 5, 279−284.
Nazeeruddin, M. K.; Grätzel, M. Nature 2013, 499, 316−319. (37) Niemann, R. G.; Kontos, A. G.; Palles, D.; Kamitsos, E. I.;
(9) Malinkiewicz, O.; Yella, A.; Lee, Y. H.; Espallargas, G. M.; Kaltzoglou, A.; Brivio, F.; Falaras, P.; Cameron, P. J. J. Phys. Chem. C
Graetzel, M.; Nazeeruddin, M. K.; Bolink, H. J. Nat. Photonics 2014, 8, 2016, 120, 2509−2519.
128−132. (38) Pistor, P.; Ruiz, A.; Cabot, A.; Izquierdo-Roca, V. Sci. Rep. 2016,
(10) Huang, J.; Yuan, Y.; Shao, Y.; Yan, Y. Nat. Rev. Mater. 2017, 2, 6, 35973.
17042. (39) Rota, L.; Lugli, P.; Elsaesser, T.; Shah, J. Phys. Rev. B: Condens.
(11) He, H.; Yu, Q.; Li, H.; Li, J.; Si, J.; Jin, Y.; Wang, N.; Wang, J.; Matter Mater. Phys. 1993, 47, 4226−4237.
He, J.; Wang, X.; Zhang, Y.; Ye, Z. Nat. Commun. 2016, 7, 10896. (40) Hunsche, S.; Heesel, H.; Ewertz, A.; Kurz, H.; Collet, J. H. Phys.
(12) Saba, M.; Cadelano, M.; Marongiu, D.; Chen, F.; Sarritzu, V.; Rev. B: Condens. Matter Mater. Phys. 1993, 48, 17818−17826.
Sestu, N.; Figus, C.; Aresti, M.; Piras, R.; Lehmann, A. G.; Cannas, C.; (41) Masumoto, Y.; Fluegel, B.; Meissner, K.; Koch, S. W.; Binder,
Musinu, A.; Quochi, F.; Mura, A.; Bongiovanni, G. Nat. Commun. R.; Paul, A.; Peyghambarian, N. J. Cryst. Growth 1992, 117, 732−737.
2014, 5, 5049. (42) Shah, J. Ultrafast Spectroscopy of Semiconductors and Semi-
(13) Piatkowski, P.; Cohen, B.; Javier Ramos, F.; Di Nunzio, M.; conductor Nanostructures; Springer Science & Business Media: New
Nazeeruddin, M. K.; Grätzel, M.; Ahmad, S.; Douhal, A. Phys. Chem. York, 2013.
Chem. Phys. 2015, 17, 14674−14684. (43) Valverde-Chávez, D. A.; Ponseca, C. S.; Stoumpos, C. C.;
(14) Zhai, Y.; Sheng, C. X.; Zhang, C.; Vardeny, Z. V. Adv. Funct. Yartsev, A.; Kanatzidis, M. G.; Sundström, V.; Cooke, D. G. Energy
Mater. 2016, 26, 1617−1627. Environ. Sci. 2015, 8, 3700−3707.
(15) Manser, J. S.; Kamat, P. V. Nat. Photonics 2014, 8, 737−743. (44) Hu, B. B.; de Souza, E. A.; Knox, W. H.; Cunningham, J. E.;
(16) Trinh, M. T.; Wu, X.; Niesner, D.; Zhu, X.-Y. J. Mater. Chem. A Nuss, M. C.; Kuznetsov, A. V.; Chuang, S. L. Phys. Rev. Lett. 1995, 74,
2015, 3, 9285−9290. 1689−1692.

18269 DOI: 10.1021/jacs.7b09508

J. Am. Chem. Soc. 2017, 139, 18262−18270
Journal of the American Chemical Society Article

(45) Bar-Ad, S.; Chemla, D. S. Mater. Sci. Eng., B 1997, 48, 83−87.
(46) Ziffer, M. E.; Mohammed, J. C.; Ginger, D. S. ACS Photonics
2016, 3, 1060−1068.
(47) Roiati, V.; Mosconi, E.; Listorti, A.; Colella, S.; Gigli, G.; De
Angelis, F. Nano Lett. 2014, 14, 2168−2174.
(48) Deschler, F.; Price, M.; Pathak, S.; Klintberg, L. E.; Jarausch, D.-
D.; Higler, R.; Hüttner, S.; Leijtens, T.; Stranks, S. D.; Snaith, H. J.;
Atatüre, M.; Phillips, R. T.; Friend, R. H. J. Phys. Chem. Lett. 2014, 5,
1421−1426.
(49) Anand, B.; Sampat, S.; Danilov, E. O.; Peng, W.; Rupich, S. M.;
Chabal, Y. J.; Gartstein, Y. N.; Malko, A. V. Phys. Rev. B: Condens.
Matter Mater. Phys. 2016, 93, 161205.
(50) Yang, Z.; Surrente, A.; Galkowski, K.; Bruyant, N.; Maude, D.
K.; Haghighirad, A. A.; Snaith, H. J.; Plochocka, P.; Nicholas, R. J. J.
Phys. Chem. Lett. 2017, 8, 1851−1855. Please also see the relevant
references in the article.
(51) Sestu, N.; Cadelano, M.; Sarritzu, V.; Chen, F.; Marongiu, D.;
Piras, R.; Mainas, M.; Quochi, F.; Saba, M.; Mura, A.; Bongiovanni, G.
J. Phys. Chem. Lett. 2015, 6, 4566−4572.
(52) Klimov, V. I. J. Phys. Chem. B 2000, 104, 6112−6123.
(53) Gdor, I.; Shapiro, A.; Yang, C.; Yanover, D.; Lifshitz, E.;
Ruhman, S. ACS Nano 2015, 9, 2138−2147.
(54) Sewall, S. L.; Franceschetti, A.; Cooney, R. R.; Zunger, A.;
Kambhampati, P. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80,
081310.
(55) Xing, G.; Mathews, N.; Sun, S.; Lim, S. S.; Lam, Y. M.; Grätzel,
M.; Mhaisalkar, S.; Sum, T. C. Science 2013, 342, 344−347.
(56) Filippetti, A.; Mattoni, A.; Caddeo, C.; Saba, M. I.; Delugas, P.
Phys. Chem. Chem. Phys. 2016, 18, 15352−15362.
(57) Yaffe, O.; Guo, Y.; Tan, L. Z.; Egger, D. A.; Hull, T.; Stoumpos,
C. C.; Zheng, F.; Heinz, T. F.; Kronik, L.; Kanatzidis, M. G.; Owen, J.
S.; Rappe, A. M.; Pimenta, M. A.; Brus, L. E. Phys. Rev. Lett. 2017, 118,
136001.
(58) Bakulin, A. A.; Selig, O.; Bakker, H. J.; Rezus, Y. L. A.; Müller,
C.; Glaser, T.; Lovrincic, R.; Sun, Z.; Chen, Z.; Walsh, A.; Frost, J. M.;
Jansen, T. L. C. J. Phys. Chem. Lett. 2015, 6, 3663−3669.
(59) Xie, L.-Q.; Zhang, T.-Y.; Chen, L.; Guo, N.; Wang, Y.; Liu, G.-
K.; Wang, J.-R.; Zhou, J.-Z.; Yan, J.-W.; Zhao, Y.-X.; Mao, B.-W.; Tian,
Z.-Q. Phys. Chem. Chem. Phys. 2016, 18, 18112−18118.
(60) Guo, Y.; Yaffe, O.; Paley, D. W.; Beecher, A. N.; Hull, T. D.;
Szpak, G.; Owen, J. S.; Brus, L. E.; Pimenta, M. A. Phys. Rev. Mater.
2017, 1, 042401.
(61) Kumawat, N. K.; Dey, A.; Kumar, A.; Gopinathan, S. P.;
Narasimhan, K. L.; Kabra, D. ACS Appl. Mater. Interfaces 2015, 7,
13119−13124.
(62) Phuong, L. Q.; Nakaike, Y.; Wakamiya, A.; Kanemitsu, Y. J. Phys.
Chem. Lett. 2016, 7, 4905−4910.

18270 DOI: 10.1021/jacs.7b09508

J. Am. Chem. Soc. 2017, 139, 18262−18270