Carrier Formation Dynamics in Prototypical Organic Solar Cells as Investigated by Transient Absorption Spectroscopy

1Graduate School of Pure and Applied Science, University of Tsukuba, Tsukuba 305-8571, Japan 2Center for Integrated Research in Fundamental Science and Engineering (CiRfSE), University of Tsukuba, Tsukuba 305-8571, Japan 3Tsukuba Research Center for Interdisciplinary Materials Science (TIMS), University of Tsukuba, Tsukuba 305-8571, Japan 4Research Center for Functional Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan


Exciton-Carrier Conversion in Organic Solar Cells.
Organic solar cells (OSCs) with bulk heterojunction (BHJ) [1][2][3][4] are promising energy conversion devices with high power conversion efficiency (PCE > 10% [5]), flexibility, and low-cost production process, for example, the roll-to-role process.The BHJ active layer, which consists of phaseseparated nanosize domains of the donor (D) and acceptor (A) materials, efficiently absorbs the solar energy and converts it to the electric energy.The BHJ layer is easily prepared by the spin-coating from an organic solvent and the appropriate thermal annealing.In some cases, an additive, for example, diiodooctane (DIO), in the organic solvent is effective in obtaining a fine domain structure [6].In the actual OSCs, the active layer is sandwiched between an Al cathode and an indium tin oxide (ITO) transparent anode.The ITO electrode is used as an optical window.
Significant feature of the OSCs is that Frenkel-type exciton with a high binding energy is stable even at room temperature, reflecting the small dielectric constant ( = 2-3) [7,8].Then, the photoexcitation creates donor exciton (D * ) in the donor domain or acceptor exciton (A * ) in the acceptor domains.The photovoltaic effect is realized by the conversion process from exciton to carrier.This is in a sharp contrast with inorganic solar cells (ISCs), in which the photoexcitation directly creates free carriers in the active layer.Figure 1 schematically shows the exciton-carrier conversion process around D/A interface.The photoirradiation directly creates excitons in the respective domains [(2) exciton formation].The excitons migrate to the D/A interface [(3) exciton migration] and dissociate into donor hole (D + ) and acceptor electron (A − ) [( 4) exciton dissociation].The exciton dissociation process is the most important issue to comprehend the photovoltaic effects in OSCs.Importantly, the exciton migration distance (∼10 nm) is very short in OSCs [9].If the domains size is much larger than the migration distance, most of the excitons recombine before they reach the D/A interface.Therefore, the nanosize domain structure of the BHJ layer is indispensable for the efficient carrier formation process.After the carrier formation, D + transfers within the D domain to reach the anode, where the hole is collected.A − transfers within the A domain to reach the cathode, where the electron is collected.
There exists a long-lasting debate whether the carriers are generated from the hot CT or relaxed CT state.In the hot CT picture, the excess energy helps to dissociate the CT states directly to the free carriers before they reach the relaxed CT states [42].In this picture, the primary kinetic competition is between the energy relaxation and the dissociation of the hot CT state.The energy relaxation process is usually several hundred femtoseconds [43].However, several experimental results on the internal quantum efficiency (IQE) [44,45] are against the hot CT picture.For example, Vandewal et al. [45] reported that IQE is irrespective of the excitation energy in the MEH-PPV/PCBM OSC.This data strongly indicates that free carrier generation is exclusively from the relaxed CT state rather than from the hot CT state.In this situation, a new spectroscopic approach is desired to deepen the understanding of the exciton dissociation.

Spectroscopic Determination of Absolute Numbers of Photogenerated Species.
The transient absorption spectroscopy is one of the powerful tools used to clarify the dynamics of the exciton, CT state, and carrier.The spectroscopy has been applied to the OSCs with BHJ active layers [42,43,[46][47][48][49][50][51][52][53][54][55][56][57][58][59][60][61]. Figure 2 shows schematic diagram of the transient absorption spectroscopy.The irradiation of the pump light pulse creates D * , A * , D + , and A − .These species cause the photoinduced absorption (PIA) below the absorption edge, as shown in the right panels of Figure 2.This PIA is monitored by the probe light pulse against delay time (t).
Here, let us introduce a method to determine the absolute numbers of the excitons and carriers against  only from the photoinduced absorption (PIA) and electrochemically induced absorption (EIA) spectra.In the latter spectroscopy, the electrochemically doped carrier causes a characteristic EIA in the infrared region.In this method, we only consider the photogenerated exciton and carrier and eliminate material dependent assumptions on the spectral interpretation as much as possible.Then, PIA ( exp ) is decomposed into the respective PIA components, that is,  D * (donor exciton),  A * (acceptor exciton), and  D + A − (carriers). D * ( A * ) should be the PIA of the donor (acceptor) neat films at the early stage after photoexcitation. D + A − should be the PIA of the blend film at the late stage, where photogenerated excitons completely disappear.We note that  D + A − is dominated by the PIA due to D + , because the profile of the late component significantly depends on the donor material and resembles that of the EIA of the donor neat film.This suggests that  A − is much weaker than  D + .The CT state inevitably coexists with excitons because its lifetime (<50 fs [41]) is very short.Therefore, it is impossible to determine the PIA ( CT ) due to the CT state without material dependent assumptions.By the spectral decomposition, the relative numbers of the excitons ( D * and  A * ) and carrier [ D + (=  A − )] can be determined against .Here, we define  photon ,  exciton , and  carrier as the numbers of absorbed photons, excitons, and carriers, respectively.Δ exciton / photon (Δ exciton is the exciton components of spectral change) is evaluated from the PIA of the blend film.Δ exciton / exciton is evaluated from the PIA of the neat film, with assuming that one absorbed photon creates one exciton.Then, the absolute number ( exciton ) of the photogenerated excitons per an absorbed photon is expressed as Similarly, Δ carrier / photon (Δ carrier is the carrier components of spectral change) is evaluated from the PIA of the blend film.On the other hand, Δ carrier / carrier is evaluated from the EIA of the neat film.Then, the absolute number ( carrier ) of the photogenerated carriers per an absorbed photon is expressed as The drawback of this method is that it does not explicitly include the CT state, which is believed to play an essential role in the carrier formation process.Nevertheless, the simple model, which excludes material dependent assumptions, has two advantages over the conventional complicated models.First, the model enables us to determine the absolute numbers of the excitons (carriers) against  only from the spectroscopic data.Secondly, the model is applicable to even unknown D/A systems, because we can experimentally determine  D * ,  A * , and  D + A − without any material dependent assumptions.
We note that the temperature effect on the carrier formation dynamics gives us a clue on the carrier formation mechanism.For example, Yonezawa et al. [60] reported that the carrier formation efficiency (Φ CF ), which is defined as the number of the photoinduced carriers per an absorbed photon, is nearly insensitive to temperature in several OSCs.This suggests that the exciton dissociation is treated by quantum-mechanical approach [62][63][64][65][66][67] and not by the socalled Marcus picture [68].In the former picture, the exciton dissociation is quantum-mechanically treated by the timeevolution of a wave function.In the latter picture, the charge separation is classically treated by the energy shift induced by displacement of the surrounding molecules.
The domain structure of the BHJ layer is complicated and its relation to the PCE value is still controversial.By means of the scanning transmission X-ray microscopy (STXM), Collins et al. [77] investigated the domain structure of PTB7/PC 71 BM blend film prepared without additive.They found that the PTB7-rich domain shows considerable fullerene mixing.By means of atomic force microscopy (AFM) coupled with plasma-ashing technique, Hedley et al. [76] observed a substructure of ∼10 nm inside the fullerene domain (∼100 nm) of the PTB7/PC 71 BM blend film prepared without additive.Such a complexity of the domain structure of the BHJ layer may prevent a true understanding of the carrier formation dynamics.In this sense, a planar heterojunction (HJ) OSC with well-defined D/A interface is suitable for detailed investigation on the carrier formation and recombination dynamics [78][79][80].For example, Moritomo et al. [80] revealed carrier density effect on the carrier recombination process in PTB7/C 70 HJ solar cell.
In this review article, we introduce a method to determine the absolute numbers of the excitons and carriers against  only from the PIA and EIA spectra.The method was applied to the prototypical OSCs, that is, (i) PTB7/PC 71 BM blend, (ii) rr-P3HT/PCBM blend, (iii) PTB7/C 70 bilayer, and (iv) SMDPPEH/PC 71 BM blend films.Figure 3 shows energy level diagram for respective D/A interfaces.Quantitative analyses clarified important aspects of the carrier formation dynamics in the OSCs.First, the late decay component of exciton does not contribute to the carrier formation process, as observed in (i) PTB7/PC 71 BM blend, (ii) rr-P3HT/PCBM blend, (iii) PTB7/C 70 bilayer, and (iv) SMDPPEH/PC 71 BM blend films.This is probably because the late component has not enough excess energy to separate into electron and hole at D/A interface.Secondly, the exciton-carrier conversion process is insensitive to temperature, as observed in (iii) PTB7/C 70 bilayer and (iv) SMDPPEH/PC 71 BM blend films.This observation, together with the fast carrier formation time in OSCs, is consistent with the hot exciton picture.

Experimental Technique
2.1.Transient Absorption Spectroscopy.Figure 4 shows a prototypical setup for the transient absorption spectroscopy.Second harmonics (=400 nm) of the regenerative amplified Ti:sapphire laser are usually used as the pump pulse.In some cases, the wavelength of the pump pulse is converted with use of an optical parametric amplifier (OPA).A white pulse, which is generated by self-phase modulation in a sapphire plate, is used as the probe pulse.The delay stage controls the delay time (t) between the pump and probe pulse.The spectra of the transmitted probe pulse are analyzed with a multichannel detector attached to a spectrometer.The differential absorption spectra (ΔOD) are defined as ΔOD = − ln( on / off ), where  on and  off are the transmitted light intensity with and without pump excitation, respectively.The photogenerated exciton and carrier cause characteristic PIA in the infrared region.

Electrochemical Differential Absorption Spectroscopy.
The electrochemical differential absorption spectra (ΔOD EC ) [61] are defined as ΔOD EC = − ln( doped / non ), where  doped and  non are the transmitted spectra of hole-doped and nondoped films, respectively.The electrochemically doped carrier causes a characteristic EIA in the infrared region.The doped carrier density ( D + ) is evaluated from the current density and doping time.The electrochemical carrier-doping was usually performed in two-pole electrochemical cell with a pair of quartz windows.The electric current is parallel to the light path.The cathode and anode are the donor neat film and a small piece of Li metal.The electrolyte is usually propylene carbonate (PC) solution containing 1 mol/l LiClO 4 .The PIA is originated from the photoinduced carriers (D + ) of PTB7 [57][58][59][60].Actually, the spectral profile is similar to that of the electrochemical differential absorption spectra of the PTB7 neat film (vide infra).Therefore, the PIA in the late stage can be used as  D + A − .We note that the photoexcitation of the BHJ layer creates the excitons, not the carriers.Therefore, the PIA in the early state (<1 ps) is overlapped by PIA due to  D * and  A * .Figure 5(b) shows ΔOD spectra of PC 71 BM neat film.The PIA due to A * is very flat and the spectral shape is independent of .Then, the PIA can be used as  A * .

PTB7/PC 71 BM Blend Film: A Prototypical Example
Figure 5(c) shows ΔOD spectra of PTB7 neat film.PIA due to D * shows broad peak around 1200 nm and the spectral shape is nearly independent of t (<10 ps).Then, the PIA (<10 ps) can be used as  D * .Now, we can decompose  exp of PTB7/PC 71 BM blend film into  D + A − ,  A * , and  D * .The spectral weights of the respective components were determined so that they minimize the following trial function: The PIA of the blend film at 3 ps was used as  D + A − .The PIA of PC 71 BM neat film at 1 ps was used as  A * .The PIA of PTB7 neat film at 1 ps was used as  D * .Figure 6 shows examples of the decomposition of the PIA spectra into  D + A − ,  A * , and  D * .Strictly speaking, the cross sections for the respective species may differ between the neat and blend films.The difference is considered to be negligible because the PIA is measured in the infrared region below the optical gap.

Absolute Numbers of the Elementary Excitations.
In order to spectroscopically evaluate  D + ,  D * , and  A * per an absorbed photon, we need the absolute intensity of the PIA per unit densities of D + , D * , and A * .The absolute intensity of the PIA due to D + is easily evaluated by the electrochemical differential absorption spectroscopy [61].Figure 6 shows examples of the electrochemical differential spectra (ΔOD EC ).The shoulder-like structure in ΔOD EC spectrum [Figure 7(a)] of rr-P3HT neat film is analogous to the PIA spectrum of rr-P3HT/PCBM blend film.Similarly, the broad peak structure in ΔOD EC spectrum [Figure 7(b)] of the PTB7 neat film is analogous to the PIA spectrum of the PTB7/PC 71 BM blend film.The absolute intensity of the PIA can be evaluated from ΔOD EC spectrum of PTB7 neat film  with considering the electrochemically doped carrier number per unit area.Concerning D * (A * ), it is reasonable to assume that one absorbed photon creates one D * (A * ) in the donor (acceptor) neat film.Then, the absolute intensity was evaluated from ΔOD spectrum of PTB7 (PC 71 BM) neat film with considering the absorption photon number per unit area.

Carrier Formation Dynamics.
Figure 8 shows  D + ,  D * , and  A * per an absorbed photon in PTB7/PC 71 BM blend film against .The solid curves are results of least-squares fittings with exponential functions.The carrier formation time ( D + = 0.3 ps) is very fast.Fast  D + is ascribed to the molecular mixing [77] as well as the nanosize domain structure [76] of the BHJ.Here, we emphasized that the excitation pulse at 400 mn was dominantly absorbed by the acceptor fullerene, not by PTB7 [see upper panels of Figures 5(a There are two possibilities to why the late component of A * does not dissociate.One possibility is that the exciton recombines before it reaches D/A interface.This scenario, however, cannot explain fast  A * (=1.5 ps) of A * , as follows.Without D/A interface, the decay channel of exciton is (i) radiative recombination, (ii) triplet exciton formation, and (iii) exciton-exciton annihilation.Among them, the former two channels cannot explain fast  A * (=1.5 ps) of A * , because typical times for these channels are on the order of nanoseconds.On the other hands,  A * of neat PC 71 BM film is 120 ps [54], which can be ascribed to the (iii) excitonexciton annihilation in bulk.The value, however, is much longer than the observed  A * (=1.5 ps).Another possibility is that the exciton dissociation efficiency decreases with .This is plausible because the excess energy of A * should decrease with .Then, late A * component has not enough excess energy to separate into electron and hole across D/A interface.Fast  A * in the blend film is probably ascribed to the extra recombination process at D/A interface, such as quench on free carrier or some structural defect.By means of a timeevolution simulation of a wave packet, Iizuka and Nakayama [67] theoretically investigated exciton dissociation at D/A interface.They demonstrated that the dissociation probability increases for the hot excitons compared with the groundstate exciton owing to their small binding energies and large diameters.The present method does not explicitly include the CT state, which is believed to play an essential role on the carrier formation process.Nevertheless, if  CT is nearly the same as  D + A − , the experimental results suggest that excess energy is needed even for the formation of the CT state.

P3HT/PCBM Blend Film: Classical but Complicated System
4.1.PIA Spectra and Analyses.As mentioned in the introduction, there already exists extensive spectroscopic investigation on rr-P3HT/PCBM blend film in literatures [32][33][34][35][36][37][38][39][40][41][42][43][46][47][48][49][50][51].Nevertheless, we reanalyzed the PIA spectra of rr-P3HT/PCBM blend film [46] and tried to determine  D + ,  D * , and  A * per an absorbed photon against .Systematic analyses of the PIA spectra in the different BHJ systems may reveal common aspects of the carrier formation dynamics in the OSCs. Figure 9(a) shows ΔOD spectra of rr-P3HT/PCBM blend film.The PIA spectra in the late stage (>10 ps) show a shoulder-like structure, which is analogous to ΔOD EC spectrum [Figure 7(a)] of rr-P3HT neat film.Therefore, the PIA in the late stage can be used as By means of the least-squares fitting,  exp were decomposed into  D + A − ,  A * , and  D * .The PIA of the blend film at 3 ps was used as  D + A − .The PIA of the PCBM neat film at 1 ps was used as  A * .The PIA of rr-P3HT neat film at 1 ps was used as  D * .Figure 10 shows examples of the decomposition of the PIA spectra of rr-P3HT/PCBM blend film into  D + A − ,  A * , and  D * .In order to spectroscopically evaluate  D + ,  D * , and  A * per an absorbed photon, we need the absolute intensity of the PIA per unit densities of D + , D * , and A * .The absolute intensity of the PIA due to D + is evaluated by the electrochemical differential spectroscopy [Figure 7(a)].The absolute intensity of the PIA due to D * (A * ) is evaluated from ΔOD spectrum of rr-P3HT (PCBM) neat film.

Carrier Formation Dynamics.
Figure 11 shows  D + ,  D * , and  A * per an absorbed photon in rr-P3HT/PCBM blend film against .The solid curves are results of least-squares fittings with exponential functions.The present analysis fails to observe the carrier formation process, because  D + monotonously decreases with decay time of 4.4 ps.One possible reason for this unexpected result is that the carrier formation time is too fast to observe.Consistently, Hwang et al. [32] proposed fast formation (<0.25 ps) of the interfacial charge-transfer states.Another possible reason may be the variation of the cross section of the carrier.The present method assumes constant cross section of the carriers.The extended electronic state of the CT state, however, may enhance the cross section as compared with that of the small polaron state (free carrier).If the cross section of the CT state is higher than that of free carrier,  D + value is overestimated in the early stage.Consistently, Takahashi et al. [86] reported intense PIA due to the CT state at -sexithiophene (6T)/C 60 interface below ∼10 ps.This observation implies high oscillator strength of the CT state.On the other hand,  D * slowly decreases with the decay time ( D * = 2.9 ps).This indicates that the late decay component of D * does not contribute to the carrier formation process, similar to the case of PTB7/PC 71 BM blend film.By means of the least-squares fitting,  exp were decomposed into  D + A − ,  A * , and  D * .The PIA of the blend film at 10 ps was used as  D + A − .The PIA of C 70 neat film at 1 ps was used as  A * .The PIA of PTB7 neat film at 1 ps was used as  D * .Figure 13 shows examples of the decomposition of the PIA spectra of PTB7/C 70 bilayer film into  D + A − ,  A * , and  D * .The absolute intensity of the PIA due to D + is evaluated by the electrochemical differential spectroscopy [Figure 7(a)].The absolute intensity of the PIA due to D * (A * ) is evaluated from ΔOD spectrum of PTB7 (C 70 ) neat film.Strictly speaking, the cross sections for the respective species may depend on temperature.In this sense, one should be careful when he discusses temperature effect on the absolute number of  D + ,  D * , and  A * .On the other hand, the temporal behaviors of  D + ,  D * , and  A * are independent of the cross section.does not contribute to the carrier formation process, similar to the case of PTB7/PC 71 BM blend film.Overall carrier formation dynamics at 80 K [Figure 14(b)] is similar to that at 300 K [Figure 14(a)].In particular, the carrier formation efficiency (=0.5 carriers/absorbed photon) at 80 K is almost the same as that at 300 K. Thus, the transient absorption spectroscopy revealed that the exciton-carrier conversion process is insensitive to temperature.We emphasize that PCE of OCS steeply decreases with decrease in temperature, because carrier transport process is significantly sensitive to temperature.This observation, together with the fast carrier formation time in OSCs [32,41], is consistent with the hot exciton picture.The low temperature effect is the elongation of carrier formation ( D + = 1.5 ps) and exciton decay ( D * = 1.8 ps and  A * = 5.1 ps) times.

Summary
We applied a new method, which determines the absolute numbers of the excitons and carriers only from the PIA and EIA spectra, to (i) PTB7/PC 71 BM blend, (ii) rr-P3HT/PCBM blend, (iii) PTB7/C 70 bilayer, and (iv) SMDPPEH/PC 71 BM blend films.The analyses revealed important common features on the carrier formation dynamics in the OSCs.First, the late decay component of exciton does not contribute to the carrier formation process, as observed in the (i) PTB7/PC 71 BM blend, (ii) rr-P3HT/PCBM blend, (iii) PTB7/C 70 bilayer, and (iv) SMDPPEH/PC 71 BM blend films.This is probably because the late component has not enough excess energy to separate into electron and hole at D/A interface.Secondly, entire carrier formation dynamics is insensitive to temperature, as observed in (iii) PTB7/C 70 bilayer and (iv) SMDPPEH/PC 71 BM blend films.This observation, together with the fast carrier formation time in OSCs [32,41], is consistent with the hot exciton picture.

2 InternationalFigure 1 :
Figure 1: Exciton-carrier conversion process at D/A interface.Molecular structures of prototypical donors and acceptors are shown.

Figure 3 :
Figure 3: Energy level diagrams of various D/A interface.

Figure 4 :
Figure 4: Schematic illustration of the experimental setup for transient absorption spectroscopy.

3. 1 .
Decomposition of the PIA Spectra.First of all, let us demonstrate how to decompose the PIA into the respective components, that is,  D + A − ,  A * , and  D * .Figure5(a)shows ΔOD spectra of the PTB7/PC 71 BM blend film.The PIA spectra in the late stage (>1 ps) show broad PIA around 1150 nm.

Figure 5 :
Figure 5: Differential absorption (ΔOD) spectra of (a) PTB7/PC 71 BM blend, (b) PC 71 BM neat, and (c) PTB7 neat films at 300 K. Top panels show optical density (OD) of the films at 300 K. Downward arrows represent the wavelength of the pump pulse.The data were replotted from [54].

Figure 6 :Figure 7 :
Figure 6: ΔOD spectra of PTB7/PC 71 BM blend film at (a) 0.3 ps and (b) 2.7 ps at 300 K. Solid curves are results of the decomposition into  D + A − ,  A * , and  D * .
) and 5(b)].Consequently, the photoexcitation only creates A * .It is interesting that exciton decay time ( A * = 1.5 ps) is much slower than  D + .This indicates that the late decay component of A * does not contribute to the carrier formation process.

Figure 8 :Figure 9 :Figure 10 :Figure 11 :Figure 12 :
Figure 8: Absolute number of carriers ( D + ), donor excitons ( D * ), and acceptor excitons ( A * ) per an absorbed photon against the delay time in PTB7/PC 71 BM blend film.Adjacent averages were plotted in  A * .The solid curves are results of the least-squares fittings with exponential functions.

Figure 13 :Figure 14 :Figure 15 :
Figure 13: ΔOD spectra of PTB7/C 70 bilayer film at 0.9 ps at (a) 300 K and (b) 80 K. Solid curves are results of the decomposition into  D + A − ,  A * , and  D * .

Figure 14 Figure 16 :
Figure 16: ΔOD spectra of SMDPPEH/PC 71 BM blend film at 0.9 ps at (a) 300 K and (b) 80 K. Solid curves are results of the decomposition into  D + A − ,  A * , and  D * .
71 BM Blend Film: Small Molecule System 6.1.PIA Spectra and Analyses.Figure 15(a)  shows ΔOD spectra of SMDPPEH/PC 71 BM blend film.At 300 K, the PIA spectra show a broad absorption band in the infrared region.The peak position shifts show a red-shift from 1100 nm at 1 ps to 1200 nm at 10 ps.The red-shift disappears above 10 ps and the spectral profile becomes independent of .Therefore, the PIA in the late stage can be used as  D + A − . A * is obtained from ΔOD spectra [Figure15(b)] of PC 71 BM neat film, while  D * is obtained from ΔOD spectra [Figure15(c)] of SMDPPEH neat film.By means of the least-squares fitting,  exp were decomposed into  D + A − ,  A * , and  D * .The PIA of the blend film at 10 ps was used as  D + A − .The PIA of PC 71 BM neat film at 1 ps was used as  A * .The PIA of SMDPPEH neat film at 1 ps was used as  D * .Figure16shows examples of the decomposition of the PIA spectra of SMDPPEH/PC 71 BM blend film into  D + A − ,  A * , and  D * .The absolute intensity of the PIA due to D * (A * ) is evaluated from ΔOD spectrum of SMDPPEH (PC 71 BM) neat film.It was difficult to obtain a quantitative electrochemical differential spectra, because the SMDPPEH molecule is solvable to the organic electrolyte.

Figure 17 InternationalFigure 17 :
Figure 17:  D + ,  D * , and  A * per an absorbed photon against the delay time in SMDPPEH/PC 71 BM blend film at (a) 300 K and (b) 80 K. Adjacent averages were plotted in  A * .The solid curves are results of the least-squares fittings with exponential functions.