Charge Generation and Recombination in High Fullerene Content Organic Bulk Heterojunction Solar Cells.

Organic bulk heterojunction solar cells with a high fullerene content (larger than 70%) have been studied in this work. The device performances of this kind of solar cell could be tuned by adjusting the blend ratio in the active layer. An appropriate amount of p-type semiconductor in the high fullerene content active blend layer is beneficial for light absorbance and exciton dissociation. The proper energy alignment between the highest occupied molecular orbital of a p-type material and an n-type fullerene derivative has a strong influence on the exciton dissociation efficiency. In addition, the mechanism of photogenerated charge recombination in the solar cells has been identified through intensity-dependent current density-voltage (J-V) measurements and results show that the mechanisms governing the recombination are crucial for solar cell performance.

Introduction

Owing to the advantages of light weight, low cost, easy processability, and compatibility with flexible substrates, organic solar cells (OSCs) have attracted extensive attention in both academic and industrial areas. Key processes determining the energy conversion ability of such solar cells are much different from those of the traditional inorganic solar cells; as a result, despite the intense research conducted in this field, many fundamental aspects of OSCs are still not fully understood.[1] Charge generation dynamics in bulk heterojunction (BHJ) solar cells is vitally important for photoelectric conversion.[2] Because light absorption from fullerene acceptors is weak in the visible spectral region, the contribution of fullerene excitons to the photocurrent has been frequently underestimated or even ignored.[3] However, fullerene excitons may make a significant contribution to free charges.[3] For example, Zhang et al.[4] have reported a fullerene-based OSC with a very low donor content and found that only 5% (in volume) of 1,1-bis-(4-bis(4-methyl-phenyl)-amino-phenyl)-cyclohexane (TAPC) was needed for the dissociation of fullerene excitons and that the photocurrent generation was not limited by the exciton diffusion or dissociation but rather by the transport of holes through the blending matrix. Yang et al.[5] have demonstrated that an internal quantum efficiency as high as 80% could be obtained by adjusting the donor–acceptor ratio to 5:95; this was an indication of a Frenkel exciton dissociation process, whereafter, the output device performance was profoundly influenced by the polymer aggregation within the active film.[6] Through femtosecond transient absorption spectroscopy, Dimitrov et al.[7] found that charge generation from PC71BM excitons primarily occurs on the 0.5 ns time scale, and light absorption by fullerenes can result in high photocurrents. Burkhard et al.[8] showed that a large amount of free carriers could be generated spontaneously at room temperature in fullerene molecules and then be extracted to an external circuit.

Herein, organic BHJ solar cells with a high fullerene content are studied and the charge generation and recombination processes are particularly investigated. Reflectance spectra measurement and photoluminescence (PL) quenching results revealed that 10% poly(3-hexylthiophene) (P3HT) was appropriate for efficient exciton photogeneration and dissociation. We found that the large open-circuit voltage (Voc) in the high fullerene content devices results from the much higher relative dielectric constant, and it is also affected by the charge recombination. By comparing the performances of P3HT-based devices with N,N′-diphenyl-N,N′-bis(3-methyl-phenyl)-[1,1′biphenyl]-4,4′-diamine (TPD) and 4,4′-bis(9-carbazolyl)-biphenyl (CBP) based devices, we found that proper energy alignment between the highest occupied molecular orbital (HOMO) levels of the p-type material and the n-type fullerene derivative is of great importance for PC71BM exciton dissociation. The dominant charge recombination mechanism in the solar cells was checked through intensity-dependent J–V measurements, and it can be concluded that nongeminate recombination is the main charge recombination mechanism in the P3HT- and TPD-based cells, whereas geminate recombination dominates the loss way in the CBP-based devices.

Results and Discussion

In the devices employed herein, P3HT and PC71BM were first employed as the p-type and n-type constituents of the BHJ layer. The photovoltaic devices were fabricated with a standard indium tin oxide (ITO)/poly(styrene sulfonate)-doped poly(ethylenedioxythiophene) (PEDOT:PSS)/BHJ/poly[(9,9-bis(3′-(N,N-dimethylamino)propyl)-2,7-fluorene)-alt-2,7-(9,9-dioctylfluorene)] (PFN)/Al structure, where PEDOT:PSS and PFN were the hole-extracting and electron-extracting layers, respectively.[9] We studied the impact of the P3HT content on the devices with a constant BHJ layer thickness of 55–60 nm (Figure S1). Current density–voltage (J–V) characteristics obtained from devices with a P3HT concentration ranging from 0 to 30% are displayed in Figure a; the corresponding performance data are reported in Table . Within this concentration range, the optimum blend ratio of P3HT to PC71BM is found to be 1:9, corresponding to a maximum solar cell efficiency of 3.07%, with a short circuit current density (Jsc) of 7.38 mA cm–2, a Voc of 0.90 V and a fill factor (FF) of 46.13%. These results are very close to the reported values in the literature.[6] To facilitate the comparison, photovoltaic parameters from all devices are presented in Figure b,c. It is easy to see that the Jsc, FF, and power conversion efficiency (PCE) values strongly depended on the P3HT content. After adding P3HT from 0 to 10%, Jsc increased from 0.16 to 7.38 mA cm–2, FF increased from 27.54 to 46.13%, and PCE increased from 0.03 to 3.07%. Further increase in the P3HT content up to 30% resulted in a slight decrease in the FF to 42.16%, whereas Jsc dramatically dropped to 3.95 mA cm–2, thus resulting in a poor PCE of 1.48%. To confirm the accuracy of these measurements, external quantum efficiency (EQE) measurements (Figure S2) were carried out; the difference between the measured Jsc and the integrated photocurrent was found to be negligible. Interestingly, the Voc parameter behaved quite differently with respect to Jsc and FF for a low P3HT concentration (0–5%); Voc increased with the P3HT weight fraction (from 0.75 to 0.88 V), whereas for a higher P3HT content (5–30%) Voc exhibited a relatively stable value.

Figure 1

Current density–voltage (J–V) characteristics (a), Jsc and FF parameters (b) and Voc and PCE (c) of BHJ solar cells as a function of the P3HT content in the BHJ layer.

Table 1

Photovoltaic Parameters of P3HT:PC71BM Cells with Different P3HT Contents (AM1.5G at 100 mW cm–2)

P3HT content (%)	Jsc (mA cm–2)	Voc (V)	FF (%)	PCEmax (PCEavea) (%)
0	0.16	0.75	27.54	0.03 (0.02)
1	2.48	0.83	35.21	0.73 (0.72)
5	7.37	0.88	42.80	2.78 (2.76)
10	7.38	0.90	46.13	3.07 (3.04)
20	5.79	0.90	43.98	2.29 (2.27)
30	3.95	0.89	42.16	1.48 (1.45)

The average PCE was obtained from more than 20 devices.

To understand the impact of a small amount of P3HT on the device performance, reflectance spectra measurements were performed to inspect the absorption of incident photons under real conditions. Figure a allows comparing the reflectance spectra of devices with different P3HT contents. As the P3HT loading ratio increased from 0 to 30%, devices showed an increased reflectance in the range above 400 nm, implying that the increasing of the P3HT content hinders light absorption in the active layer; therefore, the P3HT content influences the exciton photogeneration. Normalized PL spectra from thin films excited at 480 nm are shown in Figure b. As indicated by the absorption spectrum (Figure S3), both P3HT and PC71BM could be well excited at this wavelength. For the pure PC71BM film, the emission peak observed at about 713 nm could be attributed to the emission from PC71BM singlet excitons.[10] With the addition of a small amount of P3HT into PC71BM, the peak value of the emission at 713 nm was sensibly reduced. By further increasing the P3HT content until 10%, a steadily depressed PC71BM exciton emission is observed. This lowering of the emittance is a direct indication of the PC71BM singlet exciton dissociation. When the proportion of P3HT within the blend film is increased above 20%, a new emission peak located at about 645 nm, with a shoulder around 713 nm, is observed. As there was no obvious emission from the pure PC71BM film at 645 nm, this newly emerging peak is assigned to the emission from P3HT; this is in agreement with previous observations.[11] Concerning the re-appearance of the emission at 713 nm, it is possible that a 10% P3HT content is sufficient for ensuring the efficient dissociation of PC71BM excitons, therefore, further increasing of the P3HT content is unfavorable for the best use of fullerene excitons. Results from PL measurements suggest that unquenched excitons less than or above 10% loading are the reason for the decreased Jsc values.

Figure 2

(a) Reflectance spectra of devices for different P3HT contents. (b) PL spectra for thin films with different P3HT contents. All curves were normalized to the excitation light intensity. (c) Variation of charge mobility with the P3HT content.

Because of the close relationship between the solar cell device performance and charge carrier transport properties, the space-charge-limited current (SCLC) method was adopted to determine the charge carrier mobility.[12]Figure c depicts both electron mobility and hole mobility in devices as a function of the P3HT content. The corresponding single-carrier J–V characteristics are presented in Figure S4. Clearly, the measured electron and hole mobilities, respectively, showed a monotonic decrease and increase as the P3HT weight fraction is increased. Results from the mobility test were consistent with expectations because the increase in the P3HT content would increase hole transport pathways in the blend film and at the same time hinder the continuous motion of electrons. The hole mobility starts at a value of 1.65 × 10–7 cm2 V–1 s–1 (for 1% P3HT) and then grows up to 1.26 × 10–5 cm2 V–1 s–1 (for 30% P3HT). Obviously, the hole transport is quite inefficient in these devices, as testified by the low FF (<50%). In accordance with this idea, the FF should increase monotonically with the P3HT content. However, the steady drop in the electron mobility with an increase in the P3HT ratio is the reason for the slight decrease in the FF value in the higher P3HT loading regime (10–30%). Although the charge transport seems to be more balanced for a higher P3HT content, we deem that the value of the mobility itself has a more profound influence on the FF than having a balanced charge transport.[13]

To further clarify the origin of the low FF in high PC71BM content OSCs, voltage-dependent EQE spectra were examined. Figure shows EQE spectra from a device with 10% P3HT measured at different electrical biases. As the applied voltage turned from 1 to −2 V, the EQE value increased markedly in the entire wavelength range. However, the strong electric field dependence of the EQE is not a common phenomenon in high-efficiency BJH solar cells because photogenerated charges can be separated at the BHJ interface and then extracted to the outside circuit efficiently.[8] It is well known that Frenkel excitons in the fullerene can transform directly into free carriers; these free carriers either may undergo recombination or may be extracted to the outside circuit.[14] In our case, as PC71BM domains are very large, there is a great possibility for directly generated holes and electrons to meet and then be lost through recombination. Also, as the shape of the EQE curves does not change with the applied voltage, it could be confirmed that the nongeminate recombination is the primary carrier loss pathway. When the internal electric field in the device is enhanced (going from a positive to a negative bias), the recombination rate of free charges can be strongly reduced, leading to a higher quantum efficiency. Thus, the strong dependence of the EQE upon the applied voltage is an indication of the severe charge recombination, which is also the cause of the low FF in the devices.

Figure 3

External quantum efficiency spectra of the devices with 10% P3HT measured at different electrical biases.

In our P3HT-based high fullerene content cells, Voc is dependent on the P3HT concentration (Figure c) and is much higher than that in most reported devices (blend ratio of 1:1 and Voc is typically around 0.6 V).[15,16] This interesting phenomenon may be explained by the variation of the dielectric constant (ε) of blend layers and the charge recombination in devices because Voc in OSCs could be determined using eVoc = Eeff – Eε– Erec,[17] where Eeff is the effective bandgap, Eε describes the dielectric-related Voc loss, and Erec is the energy loss caused by charge recombination. The composition-dependent relative dielectric constant of the blend layer was studied by measuring the device geometric capacitance using a simple device structure of ITO/blend layer/Al. The relative dielectric constants of the blend layer with different P3HT contents are shown in Figure , which are derived from the capacitance–voltage measurement (the inset of Figure ).[18] It is clear that the addition of P3HT results in a small relative dielectric constant. For example, ε(3.74) of the 10% P3HT containing blend layer is higher than that of the 50% P3HT containing blend layer (ε = 2.87). As a large dielectric constant of the blend layer helps the dissociation of charge transfer excitons,[19,20] the dielectric-related Voc loss in the high-ε system is less serious (with a small Eε), and this is the main reason for the much larger Voc observed in high-PC71BM devices. When the P3HT content is less than 10%, ε increases but Voc decreases. This is because the unbalanced charge transport induces charge recombination and leads to a larger Erec value. This inference has to be confirmed by the SCLC charge mobility measurements (Figure c). Beyond that, the fact that severe charge recombination happens in extremely low P3HT content devices (<10%) could also be verified by the dark J–V characteristics because the leakage current in the reverse bias is larger in the low P3HT containing devices, as shown in Figure S5.

Figure 4

Variation of the relative dielectric constant with P3HT content. Inset: capacitance–voltage characteristics of the blend layers measured to calculate the relative dielectric constant.

To gain a more in-depth understanding of the role of the p-type material in the photoactive layer, two other p-type semiconductors, TPD and CBP, were used to fabricate high fullerene content OSCs. It should be noted that TPD and CBP are typical hole-transporting materials in organic light-emitting diodes,[21] and they have weak absorption in the visible range.[22,23] In Figure , current–voltage curves measured under illumination from devices based on P3HT, TPD, and CBP are shown. The content of P3HT, TPD, and CBP in the blend layers is 10%. To facilitate the discussion, the output characteristic curve of the device with 10% P3HT content is also presented, and the parameters from J–V characterization are summarized in Table . A comparison between devices with P3HT:PC71BM and TPD:PC71BM blend layers shows a simultaneous decrease in Jsc, Voc, and FF. However, it is worth mentioning that in the TPD-based device, the Jsc value is almost half of that measured in the P3HT-based device, and only a moderate PCE of 1.09% is achieved. This implies that, to a certain extent, the small amount of TPD could provide the appropriate function in terms of photoelectric conversion. In contrast to the P3HT-based reference cell, the CBP-based device showed very poor photovoltaic properties with a PCE of only 0.02% and a Jsc of 0.08 mA cm–2; this was even lower than what was measured in the pure PC71BM device (Table ), indicating the ineffectiveness of CBP in the high PC71BM content devices.

Figure 5

J–V curves of devices with different blend layer compositions measured under AM1.5G illumination with 100 mW cm–2.

Table 2

Parameters Extracted from the J–V Characteristics of Devices with Different p-Type Materials Measured under AM1.5G Illumination with 100 mW cm–2

blend layer	Jsc (mA cm–2)	Voc (V)	FF (%)	PCEmax (PCEavea) (%)
P3HT:PC71BM	7.38	0.90	46.13	3.06 (3.04)
TPD:PC71BM	3.65	0.82	36.55	1.09 (1.08)
CBP:PC71BM	0.080	0.73	34.07	0.02 (0.02)

The average PCE was obtained from more than 20 devices.

To understand the above phenomena, the hole transport ability in the devices was investigated through a SCLC model.[13] Current density–voltage characteristics of hole-only devices (Figure a) could be described by J = (9/8)εε0μ0V2 exp(0.89(V/E0L))1/2/L3, where ε is the dielectric constant of the polymer, ε0 is the permittivity of vacuum, μ0 is the zero-field mobility, E0 is the characteristic field, J is the current density, L is the thickness of the blended film, V = Vapp – Vbi, Vapp is the applied potential, and Vbi is the built-in potential, which results from the difference in the work function of the anode and the cathode. The hole mobilities in devices containing P3HT, TPD, and CBP are 3.74 × 10–6, 9.06 × 10–6, and 1.51 × 10–6 cm2 V–1 s–1, respectively. Obviously, the difference in charge transport ability itself could not account for the wide variation of the device performance. EQE measurements were performed, and spectra for all devices are shown in Figure b. The EQE of the P3HT-based device exhibited a maximum value of 53%, whereas the TPD-based device showed a much lower value of 31%, and the CBP-based device was very close to zero quantum efficiency in the entire wavelength range.

Figure 6

(a) J–V characteristics of hole-only devices with a structure of ITO/PEDOT:PSS/blend layer/MoO3/Al. Solid lines represent SCLC regions. (b) EQE curves of devices with different blend layer compositions.

Figure shows the molecular structures of the p-type materials employed in this work and the energy level alignment in the blend layers.[21,24] It is well known that during a typical exciton dissociation process in an organic BHJ solar cell, electrons hop from the lowest unoccupied molecular orbital (LUMO) of the p-type material to the LUMO of the n-type material, and holes stay in the p-type material. To ensure a high dissociation efficiency, the gap between the LUMO levels of the p-type material and n-type material should be large enough to overcome the exciton binding energy.[25,26] However, in the present study, as most excitons in the blend layer originate from the photoexcitation of the n-type PC71BM, the dominant exciton dissociation process is the hole transfer from the HOMO of PC71BM to the HOMO of the p-type material, and this process requires that the p-type material has a much shallower HOMO level than that of PC71BM to provide the driving force. In P3HT- and TPD-based devices, the value of ΔHOMO (HOMOp-type – HOMOn-type) is 0.9 and 0.5 eV, respectively, which are believed to be large enough for allowing for the exciton dissociation.[27−29] However, in the CBP-based device, the ΔHOMO value is −0.3 eV, which is thus very unfavorable for the generation of free charges. The lack of a driving force for the PC71BM exciton dissociation in the CBP-based device directly leads to the extremely poor photovoltaic performance.

Figure 7

Molecular structures of P3HT, TPD, and CBP and energy level alignment of these materials. Hole transfer process from the HOMO of PC71BM to the HOMO of the p-type materials is also presented. ΔHOMO values of P3HT-, TPD-, and CBP-based blends are 0.9, 0.5, and −0.3 eV, respectively.

To understand the device operation, it is important to know the charge recombination process that is responsible for losses in the solar cell. Measuring the light intensity dependence of the FF and Voc has been proved to be a powerful tool for probing dominant recombination mechanisms.[30−32] The light intensity dependence of J–V curves is presented in Figure S6, and the extracted FF and Voc are shown in Figure a,b. For P3HT- and TPD-based devices, the FF decreases with increasing light intensity. It is an indication that the nongeminate recombination process is the dominant charge recombination mechanism in these two devices because the nongeminate recombination process is heavily dependent on the charge carrier density.[33,34] The FF of the CBP-based cell is relatively stable (around 36%) within the test range, indicating that geminate recombination is the main loss channel in this device because this kind of recombination is independent of the light intensity.[28] This inference is reasonable because excitons in the CBP-based device could not be separated into free charges, just as mentioned above. The recombination mechanisms were further investigated by studying the dependence of Voc on the light intensity. Typically, when bimolecular (Langevin) recombination is the loss mechanism in a BHJ solar cell, the slope of Voc versus the natural logarithm of the light intensity is kT/e (where k is the Boltzmann constant, T is the temperature, and e is the elementary charge), whereas a 2kT/e value will be obtained when trap-assisted (Shockley–Read–Hall) recombination is the sole loss mechanism.[35,36] As shown in Figure b, Voc–light intensity curves of P3HT- and TPD-based devices could be well fitted with a linear curve. The slope for the P3HT-based device is 1.11kT/e, implying that the bimolecular recombination process dominates under open-circuit conditions. A stronger dependence of Voc on the light intensity is observed for the TPD-based device with a slope of 1.39kT/e, which is a signal of relatively severe trap-assisted recombination under open-circuit conditions.[37] This also explains why the device performance of the TPD-based device is not as good as that of P3HT-based devices. A linear correlation could not be found in the CBP:PC71BM system, which is consistent with the aforementioned result that the primary recombination mechanism is totally different from the one in the P3HT- and TPD-based devices.

Figure 8

Light intensity dependence of (a) FF and (b) Voc of the devices.

Conclusions

In this study, BHJ solar cells with a small amount of photoactive p-type materials were investigated. The variation of device performance with the P3HT loading ratio suggests that at very low P3HT content, the P3HT ratio has a profound influence on solar cell parameters. Reflectance spectra measurements indicated that the increase in the P3HT content hinders light absorption in the active layer, thus influencing the exciton photogeneration. PL quenching results revealed that 10% P3HT was enough for ensuring an efficient PC71BM exciton dissociation; further increase in P3HT content was found to be unfavorable for the best use of the fullerene exciton. Charge transport properties were investigated by the SCLC method; the inference that can be drawn is that the low FF in the devices should be ascribed to the low mobility of charge carriers; the high Voc in the high PC71BM content devices results from the higher relative dielectric constant, and it is also affected by the charge recombination. Moreover, we compared the performance of P3HT:PC71BM devices with that of TPD:PC71BM and CBP:PC71BM devices, the upshot is that to ensure efficient PC71BM exciton dissociation, the HOMO of the p-type material should be below the HOMO of the n-type material. Intensity-dependent J–V characteristics demonstrated that nongeminate recombination is the dominant charge recombination mechanism in P3HT- and TPD-based devices; a more serious trap-assisted charge recombination found in the TPD-based cell is thought to be responsible for the relatively poor device performance. By rationally choosing the p-type material in the blend, it is possible to regulate the charge recombination, thus providing good insight into the fundamental physical processes involved.

Experimental Section

Materials

The acceptor material PC71BM was purchased from Solenne BV, TPD and CBP were from Luminescence Technology Corp., PEDOT:PSS was from H.C. Starck Clevios, and PFN was synthesized according to the literature.[9] All materials were used as received without further purification.

Device Fabrication

Prepatterned ITO glass substrates were first cleaned by successive ultrasonication steps in acetone, detergent, deionized water, and isopropyl alcohol. Substrates were then treated with an oxygen plasma for 5 min, and then a 40 nm thick PEDOT:PSS layer was spin-coated on the ITO-coated glass. The substrates were subsequently left to dry at 150 °C for 15 min in air and then transferred to an N2-glovebox. The active blend layer with a thickness of 55–60 nm were prepared by spin-coating chlorobenzene solution on the resulting substrates followed by thermal annealing at 110 °C for 10 min. Then, the PFN layer was deposited by spin-casting from a 0.02% (w/v) solution in methanol at 2000 rpm for 30 s. Finally, the 100 nm Al cathode was evaporated through a shadow mask under a vacuum pressure lower than 10–4 Pa to define the device active area of 0.16 cm2. All fabricated devices were encapsulated in epoxy resin inside a glovebox.

Characterization

The PCE values were determined from the current density–voltage (J–V) characteristics determined using a Keithley 2400 source-measurement unit under AM1.5G illumination by a solar simulator. The incident light was adjusted using a set of neutral density filters (Chroma Technology), and the illumination intensity of the simulator was determined using a monocrystal silicon reference cell containing a KG-5 visible color filter. The absorption spectra were recorded with a UV–vis–near-infrared spectrophotometer (UV2550, SHIMADZU, Japan). PL measurements were carried out through a fluorescence spectrometer (FLSP920, Edinburgh, United Kingdom). The capacitance spectra were recorded using an electrochemical workstation (PGSTAT204, Metrohm Autolab).