Letter pubs.acs.org/JPCL
Time-Resolved Raman Spectroscopy of Polaron Pair Formation in Poly(3-hexylthiophene) Aggregates Timothy J. Magnanelli and Arthur E. Bragg* Department of Chemistry, Johns Hopkins University, 3400 North Charles Street, Baltimore, Maryland 21218, United States S Supporting Information *
ABSTRACT: The ultrafast formation of bound charge pairs, or polaron pairs (PPs), in mixed-order aggregates of poly(3-hexylthiophene) was investigated using femtosecond stimulated Raman spectroscopy (FSRS). Spectral dynamics in the carbon−carbon stretching region reveal a significant photoinduced depletion in steady-state features associated with lamellar-stacked, ordered polymer regions upon 500 nm photoexcitation; this is followed by the appearance of red-shifted features attributable to PPs that is delayed by a few hundred femtoseconds. PP features decay with concomitant recovery of the steadystate Raman depletion over a few picoseconds. The vibrational spectrum of the PP obtained exhibits a modest red shift (650 nm) and concomitant decrease in the GSB intensity. This behavior is a signature of the exciton annihilation process that creates PPs. As two excitations meet within the aggregate, a single highly excited excitation is generated that subsequently chargeseparates; this reduction in the number of excitations induces a fast bleach recovery in addition to the disappearance of exciton absorption at longer wavelengths as charge-separated pairs are formed. Recently, it has also been suggested that transient decay near 700 nm may be associated with an unstable, short-lived PP species.25 On longer time scales (10s− 100s of ps), the TAS signature of the PPs decays, and the GSB recovers; this process is accompanied by a red shift in the wavelength at which the spectrum crosses 0 ΔOD (605 → 640 nm) and reflects that a small fraction of longer-lived chargeseparated polarons remains when the PPs have disappeared. The fluence dependence of PP kinetics in photoexcited NPs are similar to those previously observed in films and are further described in the SI (Figure S3).14,23,29 Time-resolved Raman spectroscopy presents a new perspective on PP formation as it serves as a nuclear probe of these states as well as a selective probe of the bleaching of polymer domains associated with PP formation. Figure 2b plots the NP Raman spectrum in the CC/C−C stretching region as a function of time after 500 nm photoexcitation. Time-resolved FSRS measurements also used a 643 nm Raman excitation wavelength in order to resonantly and photoselectively interrogate the vibrational spectroscopy of PPs generated in NPs. Spectral dynamics reflect a marked decrease in the totally symmetric CC and C−C stretch intensities immediately following the actinic photoexcitation at 500 nm (1 → 2 in Figure 2b). Upon photoexcitation, the steady-state Raman signal depletes by ∼40%, which is comparable to the magnitude in GSB observed from TAS measurements; this magnitude of depletion is sensible as PP formation requires a high exciton density to occur efficiently and is induced here with highintensity laser pulses. Both the CC and C−C stretches exhibit a noticeable spectral red shift that lags depletion of ground-state features (2 → 3). The spectrum in this region recovers most of its original shape and intensity within a few picoseconds, with a smaller fraction recovering over 10s−100s of ps (3 → 4). A slight, incomplete recovery in spectral intensity by 400 ps is most likely the result of localized trapping of longer-lived charge-separated pairs (polarons) also created under high-fluence excitation, as is also evident from absorption transients shown in Figure 2a. The spectral dynamics observed via time-resolved Raman are much more subtle than what is observed via TAS; yet, trends in feature intensities and peak position make it possible to decipher the underlying kinetics. Although a mere depletion in the ground-state Raman (GSR) signal would indicate only an excitation-induced reduction in the effective ground-state population, the measurable frequency shift observed in Figure 2b indicates that two species, states, or morphologies must contribute to the Raman spectrum in this region. A frequency shift could be expected if photoexcitation at 500 nm preferentially depleted ordered versus disordered polymer regions of NPs (or vice versa), provided that the steady-state Raman spectrum is inhomogeneously broadened by variations in morphology. Although this scenario would formally involve two motifs, it cannot underlie the progression in Figure 2b for two reasons. First, the red Raman excitation wavelength used here preferentially probes the ordered lamellar-stacked regions
Figure 3. Kinetic traces obtained from time-dependent spectral measurements associated with the GSB and PP transient electronic absorption (PP TA) compared to the relative intensity depletion and frequency red shift of the time-dependent, PIR spectra.
the relative intensities of the PP TA and GSB from broad-band TAS measurements against the relative time-dependent change in the CC feature intensity and frequency shift. Both the GSB and CC depletion intensity decrease quickly during the first ∼200 fs, followed by slower intensity decay on somewhat longer time scales; this time dependence is consistent with the description of the transient spectral dynamics in Figure 2a. On the other hand, the integrated PP TA and the Raman frequency shift do not exhibit the same ultrafast component and only decay on longer time scales. Further comparison of the timedependent TAS and FSRS traces is considered within the SI using parameteric plots of ground-state polymer bleaching and transient signals from the PP state. Figures 2 and 3 demonstrate that correlations between FSRS and TAS measurements support a common kinetic description of PP formation in aggregated polymer. Furthermore, FSRS measurements provide an additional perspective on the mechanism of PP formation as the photoselectivity of the Raman measurement to morphological order of the unexcited polymer enables direct observation of the relationship between the formation of PPs and a specific morphology within the mixed-order aggregate. Figure 2b demonstrates that the delayed appearance of transient Raman features in the CC/C−C stretching region is followed by their concomitant decay and a recovery of the depletion in the Raman spectrum of the unexcited polymer aggregate. As the 643 nm Raman excitation wavelength is selectively preresonant only with ordered regions of the polymer and is resonant with absorption of the PPs, the 441
DOI: 10.1021/jz502605j J. Phys. Chem. Lett. 2015, 6, 438−445
Letter
The Journal of Physical Chemistry Letters correlation in the appearance and decay of transient PP Raman and transient GSR depletion reflects that the formation of PPs is directly linked with ordered polymer morphologies. Raman spectroscopy also provides a new perspective on the properties of transient charge-separated states of aggregated polymers. Here, we have obtained the Raman spectrum of PPs through spectral analysis of a progression of Raman spectra collected at various photoexcitation intensities. Figure 4a plots the pump-induced Raman (PIR) spectrum of RR-P3HT NPs as a function of approximate pump-pulse intensity and illustrates a decrease in the CC and C−C stretching frequencies with increasing pump power when measured 350 fs after the 500 nm (actinic) photoexcitation. Much like the time-dependent data in Figure 2b, a spectral shift is observed as the photoexcitation power is increased; this shift can be quantified against the concentration of PPs generated, as shown in Figure 4b. Here, we have specifically chosen to plot the shift against the transient absorption intensity measured simultaneously as an internal standard of PP concentration; the Raman frequency shift should correlate roughly linearly with absorbance if it arises from a population-dependent superposition of ground-state and PP features. Figure 4b illustrates that the pump-intensitydependent frequency shift observed still deviates slightly from linearity at sufficiently high PP absorbances and that the trend is best modeled with a saturation model. This saturation likely arises from experimental considerations for the three-pulse FSRS measurement that become relevant at higher pump fluences, including intensity-dependent variation in the penetration of the actinic pump relative to the region of the sample where it overlaps with the Raman excitation beam. The saturation of the Raman shift with transient absorption enables us to apply a saturation model to extract a pure PP Raman spectrum from the intensity-dependent Raman progression. Spectral variations with intensity were analyzed with the use of singular value decomposition (SVD),55 with a thorough description provided in the SI. In summary, SVD extracts the PGS spectrum, which resembles the average GSR, and DF spectrum, which is added to the PGS in linear combination to recover each intensity-dependent PIR spectrum as a function of pump fluence. SVD obtains the relative weights of these for each spectrum, and fitting them to a saturation model enables extrapolation of the PP pure transient Raman (PTR) spectrum. Trends in the PGS and DF weights versus TA intensity are plotted as Figure 4c. Among the models considered, the power saturation model gives the best residuals and most rigorous end point prediction; nonetheless, all models gave rise to qualitatively similar predicted spectra, as displayed in the SI (Figure S5). The PGS, DF, and (power saturation model) PTR spectra obtained from this analysis are compared to the average GSR and an example experimental PIR spectrum in Figure 4d. The PTR spectrum of the PP and the GSR spectrum of the polymer exhibit a number of differences. First, the symmetric CC stretching feature in the Raman spectrum of the PP extracted from SVD is red-shifted by ∼15 cm−1 relative to the groundstate/steady-state Raman spectrum of the polymer. Second, the intensities of the C−C and CC stretching bands are half that of the ground-state polymer. Third, the overall spectral shape in the C−C/CC stretching region is largely maintained in the PP state, with only a modest 2.3 cm−1 decrease in CC stretching bandwidth and a slight increase in the ratio of C−C/ CC stretch intensities from 0.34 to 0.48. Finally, there is little
Figure 4. Intensity-dependent FSRS of RR-P3HT NPs. (a) Progression of (photoexcitation) power-dependent pump-induced spectra, with fluences (mW) indicated in the legend, compared to average GSR and extracted pure transient Raman (PTR) spectra. (b) Magnitude of the CC and C−C stretch red shift as a function of PP TA intensity, fit with various nonlinear saturation models. (c) Pseudoground state (PGS) and difference function (DF) weights from SVD analysis fit with identical saturation models as in (b). (d) PGS and DF component spectra obtained from SVD analysis; the linear combination of these was used to form the PTR spectrum. The GSR and a sample pump-induced spectrum are included for comparison.
to no change in the vibrational features with lower intensities outside of the C−C/CC stretching region (see Figure 4a). Although the significant reduction in the spectral intensity for the PP compared to that for the ground-state polymer 442
DOI: 10.1021/jz502605j J. Phys. Chem. Lett. 2015, 6, 438−445
Letter
The Journal of Physical Chemistry Letters
second-harmonic bandwidth compressor/OPA pair to generate 643 nm Raman excitation pulses (∼3−4 ps). Optical delays of the photoexcitation and Raman excitation pulses relative to the white-light continuum were controlled with independent translation stages. Excitation pulses were focused to appropriate diameters to ensure complete coverage of the broad-band probe by the Raman excitation pump and both by the actinic pump. The dispersed spectrum of the probe beam was collected after the sample using a spectrograph outfitted with a CCD array detector. TAS measurements were chirp-corrected using previously described methods.43 The effective time resolution of our measurements is