** Hole transfer dynamics in polymer:PC 71 BM blends**

**2.4. Polarization sensitive dynamics: signature of different excited states**

The share of charges that are immediately created in the polymer and the portion of delayed appearance of charges after hole transfer can be distinguished with the aid of polarization sensitive pump-probe experiments, or anisotropy of PIA transients (for simplicity we will use a term “anisotropy” further on). For instance, charges that are photogenerated inside the pristine polymer and probed in the same material have high probability to retain the same polarization of transition dipole moment as that for the excitation before depolarization effects become significant50, 61, 62

. On the other hand, when
PC_{71}BM molecules are photoexcited and the polymer is probed there is no reason for the
transient dipole moments of the PC_{71}BM excitation and charge-induced (polaron) absorption
of the polymer to be correlated^{31}. The anisotropy is calculated from experimental data using
eq. 2.9 (experimental section).

The share of absorption by PC_{71}BM and polymer can be described by:

, (2.2)

and

. (2.3)

for PC71*BM (eq. 2.2) and polymer (eq. 2.3), respectively, where x is a fraction of PC*71BM by
weight (varies from 0 to 1 for blends with 0 to 100% of PC_{71}BM). If such two-component

37 model is valid, the anisotropy dependence on PC71BM fraction should strictly follow the polymer absorption fraction (see derivation in section S2.1):

_{ }^{ } , (2.4)

*where A(t) is a time-dependent anisotropy of the pristine polymer. *

Such simplified approach is only valid if PC71BM excitations are close to the interface
with polymer. PC71BM excitations, which are generated further from the interface, have to
diffuse first before hole transfer can occur. Therefore, during the first picosecond bulk
excitons of PC71BM clusters hardly contribute to the response. As a result, the effective
absorption by polymer is overestimated in equation 2.4. Therefore, the overestimation effect
was taken into account by making estimates of not as yet collected excitons as explained in
the supplementary material using eq. S2.12. Additionally, direct photoexcitation of
polymer:PC_{71}BM CT states may be possible^{50, 61-65}. Photoexcitation and probe of CT states
typically exhibits high anisotropy because the same state is photoexcited and probed50, 61, 62

and no hole transfer occurs. Therefore, photoexcitation of CT states may decrease the
effective weight fraction of PC_{71}BM in the blends when recalculating anisotropy.

Anisotropy at 0.5 ps (fig. 2.4) decreases as PC_{71}BM weight fraction increases (all
blends) indicating the increasing share of charges generated via hole transfer.

RRa-P3HT:PC_{71}BM exhibits the steepest change of anisotropy because it has the highest
PC71*BM excitation contrast k = 74.On the other hand, changes of anisotropy in the blends of *
RRe-P3HT:PC71*BM are the slowest due to the smallest k = 18. *

Anisotropy of RRa-P3HT blends (fig. 2.4) was fitted reasonably well with eq. S2.12
*with k = 74 (this value originates from linear absorption in Table 2.1) and *
*A(0.5 ps) = 0.4±0.1. There are no surprises in these blends, the value of A(0.5 ps) = 0.4 is the *
maximum possible value for randomly organized molecular systems^{66}. Anisotropy of
*MDMO-PPV blends was successfully fitted with k = 25 (from Table 2.1) and *
*A(0.5 ps) = 0.35±0.01. Slight deviation of A(0.5 ps) from 0.4 can be reasoned by a simple fact *
that initially one state is photoexcited, for instance, a CT exciton (or Frenkel exciton with a
small probability), while the probe is tuned to selective detection of charges. Moreover,
0.5 ps time-scale is long enough for the charge or energy transfer inside the polymer.

Geometric relaxation of polymer molecules and depolarization of transition dipole moment could have also initiated already even within this short time.

38

Fig. 2.4 Anisotropy dependence on PC_{71}BM content by weight for RRa-P3HT:PC_{71}BM,
MDMO-PPV:PC_{71}BM and RRe-P3HT:PC_{71}BM blends (PC_{71}BM background is subtracted). The symbols
*represent experimental data. Lines are fits using eq. S2.12. The dashed blue line represents a fit with k = *
18, assuming no formation of CT state. The solid blue line represents a fit where PC71BM weight fraction
is divided by a factor of 6, assuming the formation of CT states and 5 photons out of 6 being absorbed by
*CT states, k = 18 for RRe-P3HT blends with 0-60% PC*_{71}*BM, k = 74 for RRe-P3HT blends with 70-90% *

PC_{71}BM. The dotted blue line represents the breaking point where RRe-P3HT nanocrystals are disrupted.

Note that MDMO-PPV with 70% and 80% of PC_{71}BM are dominated by the pristine PC_{71}BM response
and, therefore, are not depicted here.

*RRe-P3HT (fig. 2.4) cannot be fitted with k = 18 (from Table 2.1). Instead, the *
anisotropy result for blends with 0-60% PC71*BM can be fitted only with k ~ 3, which *
indicates that the probability of the hole transfer process is roughly a factor of 6 lower than
expected. This observation strongly indicates that the simple two-component model in eq.

S2.12 is not sufficient to explain the excessively high anisotropy. It was reported by Drori et
*al.*^{64} that RRe-P3HT excitations below the polymer bandgap lead to a formation of the CT
state, which has a response at the energy of ~0.4 eV. The ~0.4 eV probe was used to study all
RRe-P3HT:PC71BM blends in this Chapter. Therefore, the two-component model must be
complemented by the presence of CT state, which exhibits immediate excitation and
detection (probe) of the same state. Amazingly enough, this scenario is not observed in the
blends with the other two polymers. This occurs most probably due to the fact that excitation
of RRe-P3HT:PC_{71}BM blends is closest to the polymer absorption edge (fig. S2.1) where the
absorption by CT states usually occur^{50, 61-65}. Additionally, the response of these CT states in
RRa-P3HT and MDMO-PPV blends with PC_{71}BM might have a slightly different response,

39
shifted away from our probe frequency. Note that these CT states can be formed only at the
interface between polymer and PC71BM. When PC71BM molecules start to aggregate the
fraction of the interface with polymer reduces with respect to the total volume of PC_{71}BM,
therefore, probability of excitation of CT state decreases.

A very interesting feature is the breaking point between 60% and 70% of PC_{71}BM. The
sudden drop of anisotropy is correlated with the abrupt blue-shift of the red absorption edge
of RRe-P3HT blends (fig. S2.1), which is associated with the presence of RRe-P3HT
nanocrystals26, 67, 68

. Disruption of RRe-P3HT nanocrystal formation was found to occur in
RRe-P3HT:PC_{61}BM blends at similar compositions^{69-73}. Hence, changing PC_{71}BM fraction
from 60% to 70% disrupts the formation of nanocrystals of RRe-P3HT, therefore,
nanostructure becomes more disordered and more similar to RRa-P3HT. Finally, due to the
disappearance of the red absorption edge of RRe-P3HT polymer the PC71BM excitation
contrast is enhanced and consequently anisotropy decreases abruptly.

One might argue that with a nanostructure being similar for RRa-P3HT and RRe-P3HT blends with PC71BM load 70-90% the delays of the response (fig. 2.3) should be comparable but there is a difference by almost a factor of 2. The difference of the delays of the response is a strong indication that RRe-P3HT has a different morphology as compared to RRa-P3HT even with high loads of PC71BM. The most reasonable explanation for this difference of the delays of the response is that in RRe-P3HT blends with high PC71BM loads these PC71BM molecules form large domains (see Chapter 3 and fig. 3.6), therefore, significantly fewer excitons are created at the interface and the PIA response is reduced, hence, the PC71BM to polymer spectroscopic contrast is reduced again.