Transient photoinduced absorption spectra of organic photovoltaic materials
4.1.1. PIA spectra
RRa-P3HT polymer (fig. 4.6 brown circles) was investigated for excitation of polymer (pristine RRa-P3HT) and PC71BM (RRa-P3HT:PC71BM blend with 1:9 weight ratio).
RRa-P3HT exhibits a response with two maxima at > 1 eV and 0.5 eV. The spectrum for this polymer and its blend with PC71BM was fitted with Gaussian functions (this was done for the spectra of all materials studied in this Chapter and summarized in Table 4.1. The response at
> 1 eV decays almost completely within 400 ps. In the blend with PC71BM, the > 1 eV excitation the blend has a response at a slightly different energy, 1 eV and a very similar one at 0.5 eV. The feature at 0.5 eV was assigned to CS states (charges)11 earlier (fig. 4.1 LE transitions). The disappearance of strong response originating from singlet excitons at > 1 eV in the blend indicates efficient exciton dissociation to charges. This is a common observation in all efficient OSCs. Low energy (LE) response (0.5 eV) has the same spectral shape in both the pristine polymer and the blend. The peak at 0.5 eV decays by a factor of ~3.5 in the
a) ) b)
111 pristine polymer within 400 ps. In contrast, in the blend the ~0.5 eV peak increases by 50%
within 400 ps because of PC71BM exciton diffusion followed by dissociation at RRa-P3HT:PC71BM interface that produces delayed positive charges on the polymer (see Chapters 2 and 3).
Fig. 4.6 Transient IR PIA spectra of RRa-P3HT polymer (brown circles) after photoexcitation at 474 nm and RRa-P3HT:PC71BM 1:9 blend (green triangles) after photoexcitation at 680 nm. The closed and open symbols correspond to delays of 3 ps and 400 ps, respectively. Values of error bars are explained in Experimental Methods Section.
The response at 1 eV was attributed to triplet excitons by Howard et al.67. Other reports have claimed that the response of triplet excitons is typical at higher energies above 1 eV11, 51, which is outside the measurement range of the work presented in this Chapter. In the RRa-P3HT:PC71BM blend it is very unlikely that with 90% of PC71BM an observable fraction of non-dissociated excitons would remain after 400 ps. The presence of triplet excitons can be neglected also in the pristine RRa-P3HT film because triplets should appear with a nanosecond time constant49 and survive much longer than 400 ps (i.e. microseconds51).
The HE response in the blend closely resembles the response of the pristine PC71BM (fig. 4.12). Considering the overall amplitudes of the responses of pristine PC71BM and the blend with RRa-P3HT, the HE response of the blend should be ‘contaminated’ by the response originating from pristine PC71BM. Initially (at 3 ps), the 1 eV response must be dominated by the pristine PC71BM response. However, after 400 ps, the response increases due to PC71BM exciton dissociation, whereas for pristine PC71BM the response decreases.
The increase of the response at 1 eV is due to HE transitions of charges appearing after exciton dissociation.
Table 4.1 Gaussian fits. X denotes the center position, σ – width, indexes 1, 2 and 3 – number of the peak.
X1 (eV) σ1(eV) X2(eV) σ2 (eV) X3 (eV) σ3 (eV)
RRa-P3HT 3 ps 2.5 0.54 0.53 0.19 0.14 0.026
RRa-P3HT 400 ps 2.95 0.6 0.57 0.2 0.14 0.04
RRa-P3HT:PC71BM 3 ps 0.99 0.1 0.54 0.21 0.13 0.05
RRa-P3HT:PC71BM 400 ps 0.99 0.1 0.49 0.2 0.13 0.025
PCDTBT 3 ps 1 0.16 0.43 0.18 N.A. N.A.
PCDTBT 400 ps 1 0.16 0.39 0.18 N.A. N.A.
PCDTBT:PC71BM 3 ps 1.1 0.26 0.4 0.18 N.A. N.A.
PCDTBT:PC71BM 400 ps 1.1 0.22 0.38 0.16 N.A. N.A.
PTB7 3 ps 1.25 0.38 0.39 0.14 N.A. N.A.
PTB7 400 ps 1.2 0.19 0.48 0.26 N.A. N.A.
PTB7:PC71BM 3 ps 1.2 0.21 0.3 0.21 N.A. N.A.
PTB7:PC71BM 400 ps 1.2 0.23 0.26 0.24 N.A. N.A.
PCPDTBT 3 ps 0.93 0.08 0.72 0.26 — —
PCPDTBT 400 ps 0.99 0.045 1 0.28 — —
PCPDTBT:PC71BM 3 ps 1 0.15 0.38 0.15 0.12 0.014
PCPDTBT:PC71BM 400 ps 1 0.15 0.45 0.15 0.12 0.014
PBDTTPD 3 ps 1.99 0.42 0.83 0.16 0.42 0.17
PBDTTPD 400 ps N.A. N.A. 0.99 0.16 0.348 0.18
PBDTTPD:PC71BM 3 ps 0.9 0.23 0.53 0.075 0.26 0.13
PBDTTPD:PC71BM 400 ps 0.96 0.19 0.494 0.07 0.2 0.12
BTT-DPP 3 ps 525 nm N.A. N.A. 0.89 0.24 0.33 0.1
BTT-DPP 400 ps 525 nm 1.2 0.12 0.79 0.19 0.2 0.15
BTT-DPP:PC71BM 3 ps 525 nm N.A. N.A. 0.9 0.32 0.3 0.11 BTT-DPP:PC71BM 400 ps 525 nm 1.1 0.33 0.57 0.12 0.15 0.12
BTT-DPP 3 ps 770 nm N.A. N.A. 0.92 0.29 0.3 0.12
BTT-DPP 400 ps 770 nm 1.1 0.08 0.83 0.12 0.28 0.15
BTT-DPP:PC71BM 3 ps 770 nm N.A. N.A. 0.96 0.28 0.35 0.22 BTT-DPP:PC71BM 400 ps 770 nm 1.1 0.11 0.77 0.15 0.19 0.14
PC71BM 3 ps 0.78 0.41 N.A. N.A. N.A. N.A.
PC71BM 400 ps 0.84 0.41 N.A. N.A. N.A. N.A.
113 Finally, we note that the feature at 0.15 eV was reported to be due to IR active vibrations11. The amplitude of IR active vibrations decays by a factor of ~1.5 in the pristine polymer after 400 ps and remains essentially unchanged in the blend. The decay of IR active vibrational response in pristine polymer indicates that the overall population of excited states decreased after 400 ps. In contrast, the blend produced the long-lived charges and, therefore, the decay of the IR active vibrational response is not pronounced.
PIA spectra of PCDTBT polymer (fig. 4.7) have two distinct maxima, which have essentially unchanged spectral energies (within experimental accuracy) within delays of 3-400 ps for both pristine PCDTBT and a PCDTBT:PC71BM blend. The PIA spectra were fitted with two Gaussians centered at ~1 eV and 0.4 eV with similar widths of ~0.2 eV. The pristine polymer has a response with comparable amplitudes at 3 ps for 1 eV and 0.4 eV.
After 400 ps the high energy (HE) response (~1 eV) decays by a factor of ~4. Very similar decay (factor of ~4-5) of singlet excitons within 400 ps in PCDTBT film was found by Banerji et al.44.
The response of the blend exhibits different amplitudes as compared to the pristine polymer. For instance, the response at ~1 eV in the blend initially (3 ps) has only ~60% of the amplitude in the pristine polymer. This is not surprising if at least part of this response originates from excitons (S1=>Sn transitions fig. 4.1) because the presence of PC71BM in PCDTBT polymer introduces a very efficient exciton dissociation channel68. Hence, the
~1 eV response cannot be assigned to the excitons alone. This response decays only by 30%
within 400 ps in the blend. The slow decay represents the long living excited states, in contrast to the singlet excitons, which are short-lived44. Indeed, Tong et al.12 claimed that PCDTBT charges have a response in the IR range of 1-1.5 eV both in the pristine polymer film and in the blend with PC71BM. This claim is consistent with the observation that even a pristine polymer could produce a discernible photocurrent12. The data presented in fig. 4.7 show only the very edge of the spectral range 1-1.5 eV.
Fig. 4.7 Transient IR PIA spectra of PCDTBT polymer (brown circles) and PCDTBT:PC71BM 1:3 blend (green triangles) after photoexcitation at 580 nm. The closed and open symbols correspond to delays of 3 ps and 400 ps respectively. Values of error bars are explained in Experimental Methods Section.
Based on these observations we assign the response at 1 eV to the singlet excitons (S1=>Sn transitions in fig. 4.1) and CS states (HE transitions in fig. 4.1). Such spectral overlap of excitons and CS states was reported previously in other push-pull polymers as well (i.e. PCPDTBT16-18, PTB756, 69-71). Therefore, we conclude that the pristine polymer film has both exciton- and CS state-originated response at 1 eV. The excitons in the pristine polymer are dominating the response at 1 eV, based on a similar amplitude decrease within 400 ps to the singlet exciton decay44. The blend contains much fewer excitons, hence the response of CS states must be dominating at 1 eV in this case.
The low energy (LE) response (0.4 eV) in the pristine PCDTBT film decays by a factor of ~7 at 400 ps. Interestingly, the response at ~0.4 eV in the blend at 3 ps has a ~30% higher amplitude than in the pristine polymer. This response is roughly by a factor of ~2 higher than at ~1 eV in the same blend; it decreases by a factor of ~4 over 400 ps. The association with excitonic transitions is not very likely, first of all, due to an increase of the amplitude when PC71BM is added and, secondly, the spectral feature at ~0.4 eV is usually associated with transitions from CS states in polymers6, 11, 12, 18
(fig. 4.1 LE transition).
The fast decrease of the amplitude of the response at 0.4 eV means a fast decrease of the population of the excited states. It is known from the Marcus theory72, 73 that separated
115 charges have lower recombination rates as compared to strongly bound charges in, for instance, the CT state74-78. CT states were reported to have their PIA signature at the energies similar to the CS state response30. Alternatively, CT states can dissociate into separate charges (positive on the donor material and negative on acceptor) effectively reducing the response at 0.4 eV. Therefore, we conclude that the PIA at 0.4 eV at least partially comes from the CT states (fig. 4.1 CTA transitions) that exhibit more pronounced recombination or dissociation into charges within 400 ps as compared to the 1 eV peak. There might be still some response at 0.4 eV coming from CS states (LE transitions in fig. 4.1).
The pristine PCDTBT polymer consists of alternating electron donating and accepting units. It cannot be excluded that the CT excitons (i.e. the ones with displaced mean position of the electron and the hole) are formed on the pristine polymer itself13, 79 between donor and acceptor units or intermolecularly between polymer chains. These CT excitons may be the ones producing the response at ~0.4 eV. As soon as PC71BM is added the CT excitons dissociate into CT states between the donor PCDTBT and the acceptor PC71BM due to the stronger electron affinity of PC71BM. The slightly stronger response in the blend as compared to the pristine polymer can be explained by a generation of strongly bound CT states versus CT excitons. The formation of CT states in the blend versus CT excitons in the pristine polymer is reasoned by a faster recombination in the pristine polymer – a factor of ~7 as compared to a factor of ~4 in the blend. Therefore, we can conclude that in the pristine polymer the CT exciton is formed either within electron accepting and within electron donating units that are separated with an aromatic ring or between the polymer chains. On the other hand, the inter-molecular CT state is formed in the blends where the electron is located at PC71BM and the hole at the polymer.
Both pristine PTB7 (fig. 4.8) polymer and its blend with PC71BM have responses peaking at > 1 eV and 0.3-0.4 eV. The response at > 1 eV decreases by approximately a factor of 4 within 400 ps in the pristine PTB7. As was mentioned earlier, the spectral region of > 1 eV was reported to be the signature of both excitons (fig. 4.1 S1=>Sn transitions) and CS states19, 56, 69-71
(fig. 4.1 HE transitions). The fast decrease of the amplitude already implies that singlet excitons have a dominating contribution to the response. The spectrum 400 ps after photoexcitation at > 1 eV in the pristine polymer seems to resemble that of the blend with PC71BM. This is quite expected since the blend has a higher population of CS
states than singlet excitons. For the pristine polymer, one similarly expects that at long delay the CS state response dominates due to fast relaxation of singlet excitons.
Fig. 4.8 Transient IR PIA spectra of PTB7 polymer (brown circles) and PTB7:PC71BM 1:1.5 blend (green triangles) after photoexcitation at 680 nm. The closed and open symbols correspond to delays of 3 ps and 400 ps respectively. Values of error bars are explained in Experimental Methods Section.
It is worth noting that the response of the blend has a significantly higher amplitude at
> 1 eV than pristine polymer indicating that the absorption cross section should be higher for CS states than excitons at this energy. Alternatively, the efficiency of formation of CS states in the blend could be higher than photogeneration of excitons in the pristine polymer if, for instance part of excitation on the polymer end-up as CT excitons, which do not dominate the response at ~1 eV. Moreover, the red shoulder of the maximum at > 1 eV is blue shifted in PTB7:PC71BM when comparing to PTB7. The latter observation allows concluding that CS states have a response at slightly higher energy than excitons.
Interestingly, the peak at lower energy (0.3-0.4 eV) has similar amplitudes at 3 ps both in the pristine polymer and in the blend. On the other hand, the maximum of the response at 3 ps for the pristine polymer is blue-shifted with respect to the blend. The difference of spectral position may be interpreted as the response of different excited states or the different alignment of energy levels in the pristine polymer and the blend. There is also a remarkable difference in the decrease rates of the amplitude of the responses at 0.3-0.4 eV within 400 ps:
a factor of ~20 in PTB7 as compared to a factor of ~4 in the PTB7:PC71BM blend. Moreover, LE
117 the width of the LE response is broader in the blend as compared to the pristine polymer.
Both the PC71BM and the PTB7:PC71BM blend have almost the same dielectric constant80, 81, therefore it is unlikely to be the reason for the different response between PTB7 and PTB7:PC71BM blend. The differences between the spectra of the pristine PTB7 and the blend with PC71BM make it highly improbable that the responses at low energies (0.3-0.4 eV) of the polymer and the blend have the same origin.
The response by the separated charges in the pristine PTB7 can be excluded based on the very fast decrease of the response at 0.3-0.4 eV because charges have long-lifetime in the PTB7:PC71BM blends19 It is much more reasonable to follow the same argumentation logic as with PCDTBT polymer to assign the response at 0.4 eV to CT excitons. The PTB7:PC71BM blend exhibits significant relaxation within 400 ps, although not as strong as in pristine polymer. Comparing to the HE response in the PTB7:PC71BM blend the relaxation is faster at low energies. Therefore, the dominating response at LE is unlikely to be dominated by CS states but instead should be originating mostly from CT states. An alternative explanation for the differences in the spectral response of the pristine PTB7 and the blend with PC71BM could be for instance the significant difference in molecular packing.
Different molecular arrangements could lead to differences of the localization of the CT states or polarization of the surrounding molecules to a different extent (also called self-trapping); hence, the energies of CTA transitions are different. Changes in molecular packing would be expected to induce changes in energy disorder as well, which would explain the change in the width of the LE response.
PCPDTBT (fig. 4.9) polymer exhibits broad PIA in the energy range of ~0.4-1.1 eV at 3 ps. The spectrum was fitted with two Gaussians (see Table 4.1) which indicates that there are at least two separate transitions. The transition at ~0.7 eV was fitted with significantly broader width (0.26 eV) than ~0.9 eV (0.08 eV). The maximum at ~0.7 eV was reported earlier as a signature of singlet excited states17 (fig. 4.1 S1=>Sn transitions). Our data show that the PIA amplitude at ~0.7 eV decreases dramatically (more than a factor of 10) within 400 ps, which is consistent with the excitonic response. The other peak at ~0.9 eV slightly blueshifts to ~1 eV within 400 ps. It also exhibits significant decay, but it is still clearly distinguishable and observable even at 400 ps delay.
Fig. 4.9 Transient IR PIA spectra of PCPDTBT polymer (brown circles) and PCPDTBT:PC71BM 1:3 blend (green triangles) after photoexcitation at 750 nm. The closed and open symbols correspond to delays of 3 ps and 400 ps respectively. Values of error bars are explained in Experimental Methods Section.
PCPDTBT triplet excitons were reported to have a signature at similar energy
~1 eV17, 18. The decrease of the initial response at 0.9 eV (by a factor of ~5) within 400 ps in the pristine polymer implies that quantum yield of triplets or the absorption cross section should not be high. Interestingly, at 400 ps the PIA around 1 eV of pristine PCPDTBT resembles the one of PCPDTBT:PC71BM. Therefore, the transition at 1 eV has to be of the same origin in both the pristine polymer at 400 ps and the blend with PC71BM (both 3 ps and 400 ps). In contrast, the response of pristine PCPDTBT at 0.9 eV is expected to have a different origin as compared to the response at 1 eV in the blend based on the much faster relaxation during 3-400 ps. Using the same reasoning as for the transition at 0.7 eV, the transition at 0.9 eV can be ascribed to either singlet excitons or possibly CT excitons on the same molecule.
The PCPDTBT:PC71BM blend, on the other hand, does not exhibit the 0.7 eV transition at 3 ps. Instead, the response of the blend has a second distinct PIA maximum at 0.4 eV (at 3 ps delay). Naturally, excitons that give a response at 0.7 eV are efficiently dissociated in the presence of PC71BM. The increased lifetime and amplitude of the ~1 eV transition in the blend as compared to the pristine polymer suggest that it is a signature of the HE polaron band (fig. 4.1 HE transitions) consistent with earlier reports16-18. Based on the resemblance of
119 PIA at 1 eV between pristine PCDTBT (400 ps) and the blend (3 and 400 ps) we ascribe 1 eV transitions in the pristine polymer to the HE transitions of CS states as presented in fig. 4.1.
The PIA peak at ~0.4 eV observed in the PCPDTBT:PC71BM blend is either not pronounced or absent (based on Gaussian fit, see Table 4.1) in the pristine polymer suggesting that it is unlikely to be associated with the CT excitons as was done for PCDTBT and PTB7 polymers. This notion is supported by the theoretical calculations by Wiebeler et al.82 who have predicted the response of cation at a very similar energy. The peak at ~0.4 eV decreases approximately by a factor of 4 while the peak at ~1 eV reduces only by a factor of 2. This difference indicates that transitions at 0.4 eV should not be associated exclusively with CS states while other states (e.g. CT states) contribute, too.
In support for the assumption of the presence of CT states an additional experiment was performed: the differential IR spectrum was measured for pure and iodine vapor doped PCPDTBT sample (fig. 4.9 dashed line, unfortunately staining the sample with iodine vapor did not induce any changes in the IR absorption spectra of all other polymers except RRa-P3HT). The iodine molecules are known to localize the negative charges in contrast to PC71BM molecules where electrons are delocalized over the whole molecule or aggregate of the molecules of a fullerene derivative83. These conditions allow creating a stronger binding between the positive and negative charges due to their localization next to the interface, for instance by creating CT states. Fig. 4.9 dashed line shows the 0.4 eV peak at exactly the same 0.4 eV position confirming that the LE response is dominated by strongly bound CT states.
The last feature which appears in both the PIA of PCPDTBT:PC71BM and iodine doped PCPDTBT differential IR spectrum is the peak around ~0.12 eV. The maximum at 0.12 eV closely resembles the IRAV of RRa-P3HT6, 11, 30
. Therefore, we tentatively assign this peak of PCPDTBT to IRAV modes based on its spectral similarity to RRa-P3HT.
Pristine PBDTTPD exhibits a broad and strong response at 3 ps over the entire energy range of 0.1-1.1 eV (fig. 4.10). This response was fitted with three Gaussians: the first centered at ~2 eV, the second Gaussian centered at 0.85 eV and the third maximum at
~0.4 eV. The response at 3 ps decays within 400 ps by more than a factor of 20. The dominating response at >1 eV has either broad shoulder in the range of 0.1-1 eV or strongly overlapping peaks associated with different transitions. At 400 ps the spectrum can be reproduced with two Gaussians centered around 1 eV and 0.3 eV, where the former peak has a factor of ~1.5 higher amplitude. As soon as PC71BM is added, the overall amplitude of the
spectrum at 3 ps diminishes by a factor of ~5 to ~10 comparing to the pristine PBDTTPD at 3 ps, depending on the particular energy region. Therefore, the broad dominating response of
spectrum at 3 ps diminishes by a factor of ~5 to ~10 comparing to the pristine PBDTTPD at 3 ps, depending on the particular energy region. Therefore, the broad dominating response of