Visualizing excitations at buried heterojunctions in organic semiconductor blends

TLDR: An all-optical time-resolved method to probe the local energetic landscape and electronic dynamics at interfaces, based on the Stark effect caused by electron-hole pairs photo-generated across the interface found that the electronically active sites at the polymer/fullerene interfaces in model bulk-heterojunction blends fall within the low-energy tail of the absorption spectrum.

Abstract: This work was supported by the Engineering and Physical Sciences Research Council (EPSRC) and the Winton Programme for the Physics of Sustainability. A.C.J. thanks the University of Cambridge for funding (CHESS). Synchrotron measurements were undertaken on the SAXS beamline at the Australian Synchrotron, Victoria, Australia and we acknowledge the help of N. Lal with the measurements. S.H. thanks the framework project Soltech for funding.

Figures

Figure 2: Pump-push-probe data for PCDTBT:mono-PCBM (4:1) with 500 nm pump and 526 2000 nm push pulses arriving after 4.9 ps delay. (a) Full TA map with pump and push pulses 527

Figure 3: Pump-push-probe measurements and resulting EA spectra on PCDTBT:mPCBM 537 4:1 (black) and 1:4 (red) blends. (a) EA spectra obtained from global analysis on the 538 difference TA map between pump-push-probe and pump-probe datasets with the push pulse 539 arriving 0.9 ps after initial excitation. These differential TA maps represent the pure impact 540

Figure 1: Origin of the electroabsorption signal in pump-probe and pump-push-probe 504 measurements. (a) Chemical structures of PCDTBT and PCBM fullerene acceptors. (b) 505 Schematic origin of the spectral electroabsorption (EA) features due to a Stark shift of energy 506 levels of the ground state absorption S(0) in the presence of an electric field S(E). The 507 electric field shift is exaggerated to highlight the resulting spectrum, which is flipped in sign 508 for better comparison with the feature observed in transient absorption. (c) States probed by 509

Full text

Visualising Excitations at Buried Heterojunctions
1
in Organic Semiconductor Blends
2
Andreas C. Jakowetz
1
, Marcus L. Böhm
1
, Aditya Sadhanala
1
, Sven Huettner
2
, Akshay Rao*
1
3
and Richard H. Friend*
1
4
1
Cavendish Laboratory, Department of Physics, University of Cambridge, J J Thomson
5
Avenue, Cambridge, CB3 0HE, United Kingdom
6
2
Fakultät für Biologie, Chemie und Geowissenschaften, University Bayreuth,
7
Universitätsstrasse 30, 95440 Bayreuth, Germany
8
9
e-mail: ar525@cam.ac.uk, rhf10@cam.ac.uk
10
11
12
KEYWORDS: Interface, Disorder, Charge Generation, Driving Energy, Ultrafast
13
Spectroscopy, Transient Absorption, Pump-Push, SAXS, WAXS, PDS, Polymer, Fullerene,
14
Organic Photovoltaics
15

Abstract:
16
Interfaces play a crucial role in semiconductor devices, but in many device architectures they
17
are nanostructured, disordered, and buried away from the surface of the sample. Conventional
18
optical, X-ray and photoelectron probes often fail to provide interface-specific information in
19
such systems. Here we develop an all-optical time-resolved method to probe the local
20
energetic landscape and electronic dynamics at such interfaces, based on the Stark effect
21
caused by electron-hole pairs photo-generated across the interface. Using this method, we
22
found that the electronically active sites at the polymer-fullerene interfaces in model bulk-
23
heterojunction blends fall within the low-energy tail of the absorption spectrum. This
24
suggests that these sites are highly ordered compared to the bulk of the polymer film, leading
25
to large wavefunction delocalisation and low site energies. We also detected a 100fs
26
migration of holes from higher to lower energy sites, consistent with these charges moving
27
ballistically into more ordered polymer regions. This ultrafast charge motion may be key to
28
separating electron-hole pairs into free charges against the Coulomb interaction.
29

Understanding the properties of nanoscale and disordered interfaces presents a critical
30
scientific challenge, cutting across the areas of condensed-matter physics, materials science,
31
physical chemistry, and biology. A range of techniques, such as atomic resolution electron
32
microscopy, photoelectron, and X-ray measurements, has been used to probe the properties of
33
conventional ‘ordered’ interfaces, such as those in inorganic semiconductor or magnetic
34
heterostructures.
1
Yet, these techniques have proved extremely challenging to apply directly
35
to nanoscale and disordered interfaces which are often buried away from the surface of the
36
sample. This means that conventional optical, photoelectron or X-ray techniques are often
37
swamped by signal from the ‘bulk’ of the samples and not sensitive to the interface. Interface
38
specific techniques, such as sum-frequency generation (SFG), require well defined and sharp
39
interfaces in order to generate signal, which makes them unsuited to the disordered and often
40
random morphologies of these interfaces.
2
41
42
Bulk-heterojunctions (BHJs) between organic semiconductors, which comprise an intermixed
43
blend of p- and n-type semiconductor
3–8
, provide a model disordered nanoscale interface. The
44
electronic structure and disorder at and near (<5 nm from) these interfaces controls
45
wavefunction delocalisation, charge transfer, separation and recombination efficiency and
46
thus the performance of optoelectronic devices, such as organic photovoltaics (OPVs) and
47
organic light emitting diodes (OLEDs).
9–11
A tremendous amount of work has been done to
48
understand the physical and chemical structure of these interfaces.
12,13
Yet, to date no
49
techniques exist that can report directly on the dynamics of populated electronic states at and
50
near these buried interfaces.
51
52
Here, we demonstrate an all-optical method to access information on the electronic properties
53
of such buried and disordered interfaces and their neighbouring electronic sites, and use it to
54
study model BHJ OPV polymer-fullerene blends. We utilise an ultrafast pump-push-probe
55

technique which measures the quadratic Stark effect caused by electron-hole pairs generated
56
across the interface between n- and p-type semiconductors. This “electroabsorption” signal
57
provides a unique signature of the dynamics of electronic states in the interfacial region,
58
allowing us to precisely map the local energetic landscape that the charges sample as they
59
move away from the interface. Very surprisingly, we find firstly that the local bandgap of the
60
electronically active interfacial sites is strongly redshifted compared to the bulk and secondly,
61
that charges can move from higher-energy to low-energy regions on sub 100fs timescales,
62
consistent with ballistic motion of holes.
63
64
As model system we use the polymer donor [N-11”-henicosanyl-2,7-carbazole-alt-5,5-(4’,7’-
65
di-2-thienyl-2’,1’,3’-benzothiadiazole)] (PCDTBT) which is blended with one of the Phenyl-
66
C
61
-butyric acid methyl ester derivatives mono-PC
61
BM (mPCBM), bis-PC
61
BM (bPCBM),
67
and tris-PC
61
BM (tPCBM). Weight ratios between polymer and fullerene are 1:1 across the
68
set of fullerenes and 4:1, 1:1, and 1:4 for PCDTBT:mPCBM blends. The chemical structure
69
of the materials can be found in Figure 1 (a). UV-Vis and photoluminescence (PL) spectra for
70
PCDTBT and mono-PCBM, can be found in Figure S1.1. The internal quantum efficiency of
71
optimised (1:4) PCDTBT:mPCBM blends is close to 100%
14
. The system has been
72
previously well characterised optically and structurally.
15–18
For instance several studies have
73
shown that structurally all blends consist of intermixed regions of fullerene and polymer, the
74
so-called mixed-phase.
12,13
Intercalation of fullerenes between the amorphous PCDTBT
75
chains is observed
19–21
and strong similarities to fullerenes intercalating in-between MDMO-
76
PPV are reported.
19,22
Adding more fullerene leads to complete filling of the intercalated sites
77
and eventually results in pure fullerene domains, this allows for a controlled comparison of
78
charge separation and the formation of charge transfer states through the variation of
79
fullerene content.
23
The addition of side groups to the fullerenes leads to increased disorder
80
in the packing, as has been discussed previously. This is confirmed by small-/wide-angle X-
81

ray scattering (SAXS/WAXS) spectra of the different materials and compositions, which
82
showed that mono-adduct fullerenes form the largest aggregates, while adding more side
83
groups lowers aggregate size (see Section S2). Furthermore, increasing fullerene content
84
leads to larger aggregate size, which leads to formation of larger networks within the
85
fullerene domain.
24
Increased disorder in the fullerene phase has been linked to inefficient
86
charge generation, higher charge recombination and poor device performance.
25
Thus the
87
wide tunabilty of the system via the choice and amount of fullerene added made PCDTBT an
88
ideal system to elucidate the role of structure on electronic dynamics. However, the exact
89
energetic landscape at the interface and effect of the fullerene phase on the packing of the
90
polymer and how this influences charge dynamics are difficult to quantify due to lack of
91
suitable interface specific probes.
92
93
Time-resolved optical pump-probe spectroscopies provide powerful ways to study the
94
electronic properties of such systems.
26
In these methods, a laser pulse excites the sample,
95
generating excitations such as charges and excitons. Sometime later, a probe pulse
96
interrogates the sample and measures the change in absorption (transmission) induced by the
97
pump pulse. Excited states generated by the pump pulse correlate directly to a lower ground-
98
state population and will thereby lead to a ground-state bleach (GSB) in absorption. While
99
initial excitations can contribute to the overall signal with stimulated emission (SE) due to
100
their singlet character, all excited states have a photo-induced absorption (PIA) feature, the
101
spectral shape of which depends on the material and the nature of the excited state.
102
Importantly, pump-probe methods provide information on the specific site/chromophore on
103
which the excitation is when probed, but not on the surrounding sites. Furthermore, these
104
techniques do not provide any interface specificity, rather they provide information of
105
whichever site/chromophore the excitation is located on.
106
107

Frequently asked questions

Q1. What are the contributions in "Visualising excitations at buried heterojunctions" ?

51 52 Here, the authors demonstrate an all-optical method to access information on the electronic properties 53 of such buried and disordered interfaces and their neighbouring electronic sites, and use it to 54 study model BHJ OPV polymer-fullerene blends. The authors utilise an ultrafast pump-push-probe 55 technique which measures the quadratic Stark effect caused by electron-hole pairs generated 56 across the interface between nand p-type semiconductors. This “ electroabsorption ” signal 57 provides a unique signature of the dynamics of electronic states in the interfacial region, 58 allowing us to precisely map the local energetic landscape that the charges sample as they 59 move away from the interface. 12,13 Intercalation of fullerenes between the amorphous PCDTBT 75 chains is observed 19–21 and strong similarities to fullerenes intercalating in-between MDMO76 PPV are reported. 106 107 Turning to the case of interfaces, the authors find that pump-probe spectroscopy can provide 108 additional information. 11,28–30 Recently, the authors have demonstrated 121 that this EA can be tracked as a function of time at BHJ interfaces. The observed EA signal resembles the first derivative of the absorption 135 spectra of these molecules ( see more details in the SI, Section S4 ), as reported in the 136 literature on TA measurements 11,29,30 and steady-state EA experiments 31. Thus, the EA 138 signal could provide us a window on the energetic landscape experienced by charges as they 139 move away from the interface. 143 144 the authors present here a new pump-push-probe electroabsorption ( PPP ) technique, which uses a 145 third ‘ push ’ pulse in order to obtain clean EA signatures on more localised areas. The authors follow 148 this with a time-delayed ‘ push ’ pulse, at 2000nm ( 0. 62eV ), which is targeted in the low149 energy PIA of the hole-polarons. Importantly, as the authors report here, by moving the 157 hole they change the electric field distribution caused by the e-h pair, as illustrated in Figure 158 1 ( d ). As illustrated in Figure 1 ( e ), this EA signal reports only on the small subset of 163 molecules brought into, or removed from, the influence of the electric field by the movement 164 of the charge following the push pulse. For the sake of clarity, in the following text the authors focus their investigation on 4:1 and 1:4 for 174 PCDTBT: mPCBM blends. The 1:4 blend provides best solar cell operation, with quantum 175 yields for charge collection close to 100 % 14 and the 4:1 blend provides a comparison as a 176 system with efficient early time electron-hole pair generation but inefficient long-range 177 charge separation. Data for others blends, which show similar trends to those discussed in the main text, 179 are provided in the SI ( see Section S3. 11 194 Thus, the authors find that the push pulse gives rise to an EA signal, as described above, which 195 contains information on the molecules near the interface. The authors note that spectra taken from the 196 raw data show the same trend ( see Figure S3. 5 ). Figure 3 ( c ) shows 208 the absorption spectra of the neat polymer film measured with photothermal deflection 209 spectroscopy ( PDS ), which provides several orders of magnitude in detection sensitivity in 210 absorption and is not distorted by reflection and scattering. The authors consider that the very low energy of 215 these electronically active interfacial sites must be related to highly planar and ordered 216 polymer backbones, leading to larger wavefunction delocalisation and low bandgaps. This 217 suggests that in all the blends studied here, the electronic dynamics at the interface are 218 controlled by a very small fraction of sites ( which are not representative of the bulk film ) 219 which are highly ordered, planar, and of lower energy than the bulk of the polymer film. This ordering effect has been 228 reported in the literature for the amorphous polymers PCDTBT and MDMO-PPV upon 229 fullerene addition. Very surprisingly, the authors find firstly that the local bandgap of the 60 electronically active interfacial sites is strongly redshifted compared to the bulk and secondly, 61 that charges can move from higher-energy to low-energy regions on sub 100fs timescales, 62 consistent with ballistic motion of holes. Furthermore, increasing fullerene content 84 leads to larger aggregate size, which leads to formation of larger networks within the 85 fullerene domain. Furthermore, these 104 techniques do not provide any interface specificity, rather they provide information of 105 whichever site/chromophore the excitation is located on. 11 Furthermore, using an EA signal, it has been shown that also 128 polarons in a pBTTT: PCBM system can move on timescales < 1ps between different phases 129 of the BHJ, i. e. from a co-crystal region to domains of pure pBTTT. 16 Such signals also contain information from all the molecules influenced by 142 the electric field, and this reduces the spatial resolution of the information the authors can obtain. The authors 204 observe that for both blends the interfacial absorption is significantly red-shifted compared to 205 the bulk of the polymer film ( green dashed line, integration of CW EA of a 206 PCDTBT: mPCBM film ).