Unravelling the role of electron–hole pair spin in exciton dissociation in squaraine-based organic solar cells by magneto-photocurrent measurements

M. Klein *ab, S. Majumdar cd, P. Zassowski e and W. Stampor a
aDepartment of Physics of Electronic Phenomena, Faculty of Applied Physics and Mathematics, Gdansk University of Technology, Narutowicza 11/12, 80-233 Gdansk, Poland. E-mail: mklein@mif.pg.gda.pl
bDistributed Energy Department, The Szewalski Institute of Fluid-Flow Machinery, Polish Academy of Sciences, Fiszera 14, 80-231 Gdansk, Poland
cNanoSpin, Department of Applied Physics, Aalto University School of Science, P.O. Box, 15100, FI-00076 Aalto, Finland
dWihuri Physical Laboratory, Department of Physics and Astronomy, University of Turku, FI-20014 Turku, Finland
eFaculty of Chemistry, Silesian University of Technology, Strzody 9, 44-100 Gliwice, Poland

Received 5th November 2017 , Accepted 4th December 2017

First published on 5th December 2017


A high absorption coefficient and narrow absorption bands in squaraine (SQ) dyes have resulted in rapidly growing interest in them as a donor material in photovoltaic devices. The exciton dissociation process in organic systems proceeds via a multistep mechanism where the electron–hole pairs (charge transfer states) involved in the current generation process determine the recombination losses and subsequently limit the overall performance of organic solar cells. In this work, these basic electronic processes are investigated by magneto-photocurrent measurements (MPC, the photocurrent change induced by the external magnetic field) of SQ:PC60BM bulk-heterojunction solar cells with varying electron acceptor concentrations under magnetic fields up to 9 T and at different temperatures. Under a weak external magnetic field, the change in photocurrent is due to electron and hole (e–h) pairs that experience a modulating hyperfine interaction associated with nuclear (mainly proton) magnetic moment, while in strong magnetic fields the photocurrent is affected by the Δg mechanism with spin dephasing due to different Lande factors of the electron and hole entities (Δg ≈ 10−3). To consistently interpret the amplitudes and lineshapes of the MPC signals at various temperatures, charge carrier hopping in a disordered environment competing with the magnetic dipole spin precession is proposed. The requirements for efficient small-molecular weight organic:fullerene bulk-heterojunction solar cells are briefly discussed.


A Introduction

Over the last few years, the interest in squaraine (SQ) dyes as a donor material for photovoltaic (PV) applications has grown rapidly.1,2 It is associated with the unique photophysical properties of these organic compounds: a high absorption coefficient (approximately 105 cm−1) and narrow absorption bands only in the visible-near infrared region (from ∼550 nm even up to ∼1000 nm).3,4 Due to relatively simple synthesis routes, various derivatives of SQ dyes have been developed and found to be applicable in new generation PV technologies: dye-sensitized solar cells (DSSCs), organic photovoltaic (OPV) devices with both planar (PHJ) and bulk-heterojunction (BHJ) architectures and perovskite solar cells (PSCs).1 The last decade has observed a sharp increase in the photoconversion efficiency (PCE) of these devices (5.9% and 6.1% for SQ[thin space (1/6-em)]:[thin space (1/6-em)]fullerene in PHJ5 and BHJ6 architecture, respectively, 8.3% for SQ/polymer/fullerene tandem solar cells7 and over 10% for quaternary organic solar cells2) which indicates that SQ:fullerene based OPV cells have significant potential for future commercial applications. Moreover, squaraine has been considered as a possible material for spintronic applications.8 Rather short exciton diffusion length (LD ≤ 2 nm9) and low charge carrier mobility in SQ thin films, in comparison to those in PC70BM (phenyl-C70-butyric acid methyl ester, LD = 20 to 40 nm10), implied promise for use of SQ[thin space (1/6-em)]:[thin space (1/6-em)]fullerene blends with the compositional ratio strongly favoring fullerene (best photoconversion efficiencies, 5.5%, obtained for 1[thin space (1/6-em)]:[thin space (1/6-em)]6 SQ to fullerene weight ratio).10 Even though the observed low fill factors (FF) for current–voltage characteristics suggest the process of exciton dissociation/carrier pair recombination to be field-dependent,11 the authors of ref. 12 argue that in such systems with poor transport properties (high internal resistance and short length of exciton diffusion in SQ) low charge collection governs the cell performance.

Nowadays, further improvement in the performance of organic solar cells can be achieved by applying the following strategies: (i) tuning or broadening the absorption band of photoactive materials for better matching with the solar spectrum;1 (ii) employing new device architectures, especially tandem solar cells,13 maximizing the open circuit voltage (Voc) by appropriate energy level alignment, namely, a charge transfer level (in donor–acceptor systems) should lie close to a singlet state of the donor (S1);14 (iii) enhancing photocurrent densities by singlet fission based strategies15,16 and (iv) enabling the suppression of electron–hole recombination by taking into account the interplay between spin, energetics and delocalization of electronic excitations.17 The primary step of photocurrent generation in organic solar cells is the dissociation of photogenerated excitons by charge transfer across the donor–acceptor interface leading to the formation of bound interfacial charge transfer (CT) states where an electron and a hole are located on separate molecules, an acceptor and a donor, respectively. Such bound charge pairs can dissociate into free carriers generating a photocurrent or restore the ground state by means of geminate recombination.17,18 Due to the larger separation distance of electron–hole (e–h) pairs, compared to molecular excitons, relatively weak electrostatic electron exchange interactions result in CT states of singlet 1(CT) and triplet 3(CT) spin character, almost degenerate in energy. Therefore, the role of spin of electronic excitations is essential when considering the possible pathways for both photocurrent generation and recombination losses in photovoltaic devices.17–19

Recently, magnetic field effect (MFE) technique has been recognized as a powerful tool for studying spin-dependent generation and recombination processes of spin-pair species in organic semiconductors or polymer based solar cells and light emitting diodes.19–27 In state-of-the-art polymer donor:PC60BM devices, depending on the PC60BM ([6,6]-phenyl-C60-butyric acid methyl ester) concentration and external magnetic field strength, various MFEs on a photocurrent (called magneto-photocurrent, MPC) have been observed. For pristine P3HT (poly(3-hexylthiophene)) or its blend with a low PC60BM concentration (<1 wt%) in a low magnetic field (a few millitesla) the positive component of MPC is related to the hyperfine interaction modulation (HFM) in (e–h) pairs while the negative component at a higher magnetic field (tens of millitesla) reflects exciton-charge reactions (T-q model) occurring in the triplet excitonic states.21 Similar effects have been reported for blends with poly(2-methoxy-5-(3,7-dimethyloctyloxy)-1,4-phenylenevinylene) (MDMO-PPV) as a donor.22 Intermediate concentrations of PC60BM (30–60 wt%) lead to the formation of CT states at the donor–acceptor interface22,23 whereas for a high PC60BM content (over 70%) strong phase separation occurs and only a negative MFE at a low magnetic field, outlined as a bipolaron (BP) mechanism, is present.22 In a previous work,21 MPC at a high magnetic field (ca. 1 T) and intermediate PC60BM content is reported to be related to the CT states without the specification of the exact mechanism of magnetic field effects. However, a few other reports22,23 suggested that MPC is associated with the dephasing of spin magnetic dipoles due to different values of the Lande g factor (Δg ∼ 10−3) for electron and hole entities forming CT states, referred to as the Δg mechanism. In our earlier work28 we examined MFEs in dye-sensitized solar cells with SQ2 adsorbed on a TiO2 semiconductor where small negative MPC signals already observed under a weak external magnetic field were ascribed to the Δg mechanism due to the relatively low g value of the Ti3+ electron center in the semiconductor material.

In the present work, we focus on the exciton dissociation and charge carrier recombination processes in SQ:PC60BM organic solar cells based on the studies of the magnetic field effects on photocurrent. We address the role of e–h pair states and CT states in OPV cells with various PC60BM concentrations presenting a detailed photocurrent generation mechanism for both single-layer and bulk-heterojunction solar cells. Depending on the temperature and active layer content, the different components operating in the ultrasmall (<3 mT), hyperfine (<10 mT), fine (<100 mT) and high (of the order of several tesla) magnetic field range contribute to the overall MPC signal. To the best of our knowledge, we report here for the first time on the magnetic field effects in small-molecular weight organic:fullerene bulk-heterojunction solar cells and unravel the mechanism underlying these effects in a wide range of magnetic field strengths.

B Experimental

Materials

SQ2 and PC60BM were purchased from Solaronix and Lumtec, respectively, and were used as received without any additional purification.
Device fabrication. SQ:PC60BM devices were fabricated on etched ITO coated glass substrates (resistivity = 100 Ω sq−1) which were cleaned sequentially using acetone, ethanol and deionized (DI) water for 10 min each in an ultrasonic bath and then dried under a stream of dry nitrogen followed by ozone treatment for 15 min. Afterwards the substrates and materials were transferred into a nitrogen-filled glove box ([O2] < 1 ppm, [H2O] < 1 ppm). SQ2 and PC60BM were separately dissolved in anhydrous chlorobenzene (99.8%, Aldrich) at 60 °C overnight and were then mixed in order to prepare appropriate SQ2 to PC60BM weight ratios. The solutions were heated at 60 °C for 2 h just prior to spin coating. The ITO glass substrates were then transferred into the vacuum system connected directly with the glove box and a MoOx layer of 8 nm thickness was thermally evaporated at a base pressure of ∼10−6 Torr. Afterwards SQ, SQ[thin space (1/6-em)]:[thin space (1/6-em)]PCBM (1[thin space (1/6-em)]:[thin space (1/6-em)]0.1%), SQ[thin space (1/6-em)]:[thin space (1/6-em)]PCBM (1[thin space (1/6-em)]:[thin space (1/6-em)]10%), SQ[thin space (1/6-em)]:[thin space (1/6-em)]PCBM (1[thin space (1/6-em)]:[thin space (1/6-em)]6) thin films were spin coated from 10, 10, 11 and 42 mg mL−1 solutions at rates of 1500, 1500, 1500 and 1000 rpm for 90 s giving active layers of thickness around 20, 20, 23 and 75 nm (as determined by using a Tencor Alpha Step 500 Profilometer), respectively. The films were then annealed at 110 °C for 10 min. The devices were completed by thermally evaporating a cathode, consisting of 8 Å LiF and 60 nm Al, through a shadow mask in a vacuum system with a base pressure of ∼10−6 Torr.
Photovoltaic performance. Photocurrent density–photovoltage (JV) curves were recorded under 100 mW cm−2 AM 1.5 G illumination conditions in ambient air on a Keithley 2400 Sourcemeter.
Absorption spectroscopy. Active layers for UV-Vis absorption and thickness measurements were deposited onto microscopic glass slides in a similar manner to the device preparation. Absorbance spectra were recorded on a Lambda 35 UV-Vis spectrophotometer (Perkin-Elmer) applying appropriate background subtraction.
Photoluminescence spectroscopy. Phosphorescence measurements of SQ2 were performed in a homemade setup consisting of a helium cryostat and a 0.3 m Czerny–Turner spectrograph (SR303i, Andor) equipped with an ICCD camera (DH740, Andor). Samples were excited with a HeNe laser (632.8 nm center wavelength and 2 mW output power). The photoluminescence signal was collected perpendicular to the sample surface, using a quartz lens, and focused on the entrance of an optical fiber. Appropriate optical filters were used to block the excitation light. To enhance the phosphorescence by an external heavy atom (bromine) effect a saturated solution of SQ2 in 1,2-dibromoethane was cooled down to 25 K. Absorbance and fluorescence spectra of 10−5 M SQ2 solution in 1,2-dibromoethane were measured at room temperature using a Lambda 35 UV-Vis spectrophotometer (Perkin-Elmer) and a LS 55 fluorescence spectrometer (Perkin-Elmer), respectively.
EPR measurements. EPR measurements of a radical cation of SQ2 were performed using a JEOL JES-FA200, X-band CW-EPR (continuous-wave electron paramagnetic resonance) spectrometer, operating at 100 kHz field modulation, coupled with an Autolab PGSTAT100 potentiostat. Measurements were carried out in a glass cell narrowed at the bottom, to provide proper conditions for recording EPR spectra, equipped with a Pt wire working electrode, an Ag wire pseudoreference electrode (calibrated vs. Fc/Fc+) and a Pt coil counter electrode. Electrolysis was carried out at the potential corresponding to the first oxidation potential (see Fig. S2, ESI). The g factor of the generated radical cations was determined by comparison with a JEOL internal standard (Mn(II) salt), knowing that its third hyperfine line has a g factor of 2.03324.
AFM measurements. Samples for atomic force microscopy (Veeco Dimension 5000) operated in tapping mode were prepared in the same way as for absorption spectroscopy.
Magnetic field effect measurements. For MFE measurements, devices were transferred through a nitrogen filled container to a Quantum Design Physical Property Measurement System (PPMS) equipped with a superconducting magnet and kept in a vacuum (<1 mTorr). The devices were driven at a constant bias voltage U = 0.1 V using a Keithley 6487 Picoammeter/Voltage Source and were illuminated using a L658P050 laser diode (658 nm center wavelength and 50 mW output power, Thorlabs) and a few meter long optical fiber introduced into the PPMS chamber. An incident photon flux of approx. 1016 cm−2 s−1 was achieved.
Energy levels. The singlet HOMO (highest occupied molecular orbital) and LUMO (lowest unoccupied molecular orbital) energy levels of the SQ2 molecule were taken from ref. 29 while its triplet level was estimated from the maximum of the phosphorescence emission (cf. Fig. S1, ESI). Energy levels of PC60BM were taken from ref. 30. The upper limit of the CT state energy levels (ECT) was calculated from the following formula ECT ≈ |EALUMO| − |EDHOMO| = 1.1 eV.

C Results and discussion

In this study, we used single-layer solar cells, with a squaraine active layer (Fig. 1a), and bulk-heterojunction solar cells (Fig. 1b) with blends of SQ (acting as an electron donor and hole conducting material) and PC60BM (acting as an electron acceptor), with various wt% acceptor to donor ratios. The chemical structures of the materials applied are shown in Fig. 1c and d while the absorption spectra of the spin-coated films of pristine SQ, PC60BM and the SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM (1[thin space (1/6-em)]:[thin space (1/6-em)]6 wt%) blend are depicted in Fig. 1f. Fig. 1e displays the photocurrent density-photovoltage curves of OPV with the SQ and SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM (1[thin space (1/6-em)]:[thin space (1/6-em)]6) active layer under standard simulated illumination conditions (100 mW cm−2, AM 1.5). A great enhancement of photocurrent density (Jsc, from 1.1 to 7.3 mA cm−2) and photoconversion efficiencies (PCE of 0.13% and 1.28% for SQ and SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM (1[thin space (1/6-em)]:[thin space (1/6-em)]6), respectively) is clearly seen, which compares well with SQ2/C60 planar-heterojunction solar cells reported previously for the same type of squaraine molecule.31 The lower value of Voc for single-layer than for bulk-heterojunction solar cells is likely induced by the low shunt resistance of the thin SQ film subject to possible internal short-circuits. It should be noted that the present OPV devices were not optimized for the best performance being fabricated without any carrier blocking buffer layers (that significantly enhance the cell performance up to 6%6,10) to track and unravel MFEs on as simple as possible photovoltaic systems.
image file: c7tc05033j-f1.tif
Fig. 1 Structures of a single-layer (a) and a bulk-heterojunction (b) solar cell; chemical structures of squaraine dye SQ2 (c) and fullerene derivative PC60BM (d); the photocurrent density-photovoltage curves for solar cells: with a SQ (solid line) and SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM 1[thin space (1/6-em)]:[thin space (1/6-em)]6 weight ratio (dashed line) active layer (e); absorption spectra of SQ (solid line), PC60BM (dashed line) and blend (dotted line) of both materials with SQ to PC60BM 1[thin space (1/6-em)]:[thin space (1/6-em)]6 weight ratio (f).

In the following sections, the MFE technique is applied to examine the exciton dissociation and charge carrier recombination processes in systems under investigation. The MPC(B) response is given by

 
image file: c7tc05033j-t1.tif(1)
where the respective terms represent the photocurrent with and without the magnetic field (j(B) and j(0), respectively). For identifying the spin-mixing mechanism involved in the origin of MPC signals the data points have been fitted with a single-Lorentzian or a double-Lorentzian function having the form of
 
MPC = ALFEB2/(B2 + BLFE2)(2)
and
 
MPC = ALFEB2/(B2 + BLFE2) + AHFEB2/(B2 + BHFE2),(3)
respectively. In the relevant components of formulas (2) and (3) representing the low-field (LFE) and high-field (HFE) effects, ALFE and AHFE parameters denote the MPC signal magnitudes for B → ∞ whereas BLFE and BHFE determine the half width (B1/2) at half signal maximum (HWHM).28,32 Finally, we discuss the possible routes of photophysical processes responsible for the observed MFEs.

MFEs in squaraine single-layer solar cells

The magnetic field effects on photocurrent (the MPC signal) recorded as a function of the external magnetic field strength for two different temperatures in a sandwiched configuration, ITO/MoOx/SQ/LiF/Al, for the low-field and the ultrasmall-field regime have been depicted in Fig. 2a and b, respectively. At room temperature (290 K), the MPC signal saturates on a magnetic field scale characteristic of hyperfine interactions (BLFE = 4 mT) while at lower temperature (200 K), besides the low-field component (BLFE = 3 mT), the medium-field component (BMFE = 30 mT) also appears (Fig. 2a). Thus, at 290 K, according to the electron–hole pair (EHP) model,20,33–36 the external magnetic field suppresses the spin-mixing occurring at a hyperfine-field scale and consequently increases the singlet, 1(e–h), to triplet, 3(e–h), electron–hole pair (polaron pair) population ratio in a squaraine molecule. At reduced temperature, gradually increased population of triplets begins to play a more important role due to the deactivation of radiationless decay pathways. Therefore, at 200 K, besides the low-field effect (induced by HFM), another mechanism likely associated with the fine structure modulation (FSM) becomes active at a higher magnetic field. When the external magnetic field competes with the internal (fine) magnetic field of electronic spin origin, it starts to modulate the molecular triplet (T) zero-field splitting (ZFS) and thus changes the free carrier mobility. This triplet-charge interaction (T-q mechanism), proposed as the origin of magnetoresistance in organic semiconductors (organic magnetoresistance, OMAR) by Cox et al.37 and called the trion model, has been discussed in our previous work32 (see also ref. 38). Accordingly, the doublet and quartet trions are formed by the interaction between spin-1/2 free charge carriers and spin-1 triplet molecular excitons, T, (presumably trapped in defect sites of an organic solid) populated in the SQ layer. An essential condition for the functionality of this model is that the trions operate as free-carrier-capturing centers hindering the carrier mobility. As a consequence, the overall process may be interpreted as a scattering of free carriers on the triplet states. Since the recombination process of the doublet trions is spin-allowed and the recombination of the quartet trions is spin-forbidden, the lifetime of the doublet trions is much shorter than that of the quartet ones. Therefore, the quartet trions are more efficient in capturing free carriers and hence, in reducing the photocurrent. Importantly, in the FSM-scale magnetic field of several-tens-of-mT, the population of the quartet trions is reduced whereas the population of the doublet trions increases. This leads to the lower contribution of the quartet trions and, simultaneously, higher contribution of the doublet trions to the overall scattering process. One, therefore, should observe the increase in photoconductivity of an organic solid due to the increasing mobility of charge carriers which is the case in Fig. 2a. Another issue which we would like to explain is the amplitude enhancement of the low-field component due to temperature reduction. Nevertheless, first we shall note that the external magnetic field can change the intersystem crossing rate in electron–hole pair intermolecular states (ISCeh) but has little influence on this rate in intramolecular excitonic states (ISCST) with a larger singlet–triplet splitting energy.39 Therefore, on the one hand an increase in the amplitude of the MPC signal originated from carrier scattering on molecular triplets can be induced by the low magnetic field associated with the HFM mechanism, namely, the external magnetic field operating on the hyperfine scale reduces the population of T excitons formed from 3(e–h) states thereby enhancing the mobility of free charge carriers. However, on the other hand another explanation of the amplitude change, which is in good agreement with the other results presented later in the text, is also possible. If we assume that the charge carrier hopping frequency (ωhop) is on the order of hyperfine precession frequency (ωhf), i.e. ωhf/ωhop ≈ 1 (intermediate-hopping regime), then reducing the temperature, and thereby also reducing the hopping frequency, will be reflected in an increase in the MPC signal amplitude and a decrease in its linewidth.40 Indeed, a slight change in the width of a low-field Lorentz component with temperature reduction, from BLFE = 4 mT to BLFE = 3 mT, has been observed. Nevertheless, it may be also associated with the reduction of the broadly distributed decay times of (e–h) pairs in such disordered materials as those used in OPV cells.23
image file: c7tc05033j-f2.tif
Fig. 2 The magnetic field effect on the photocurrent for single-layer solar cells. The MPC signal as a function of magnetic field strength for two different temperatures: 200 K (circles) and 290 K (squares) in the low (0–140 mT) and ultrasmall (0–3 mT) field range are displayed in parts (a) and (b), respectively. The MPC signal versus: forward (negative) and reverse (positive) bias voltage (c), and photon flux (d) at 20 mT-magnetic field and 290 K temperature are also shown. The solid lines in part (a) represent the best fit according to the single- and double-Lorentzian functions for 290 and 200 K, respectively. The solid line in part (b) is a guide to the eye.

Let us now take a closer look at the 0–3 mT magnetic field strength range (Fig. 2b). The ultrasmall magnetic field effect (USMFE), with the opposite sign to the low-field effect can be easily recognized. This intrinsic effect in OMAR according to the semiclassical approach proposed by Koopmans and co-workers40 originates from the competition between spin mixing and exciton formation for intermediate-hopping rates. On the other hand, to investigate the observed MPC line shapes, Vardeny and co-workers41 applied a fully quantum-mechanical approach in which the polaron pair spin Hamiltonian includes the hyperfine interaction between each of the polaron pair constituents and one or more strongly coupled neighboring nuclei. Therefore, the existing singlet–triplet (S–T) level-crossing (LC) of (e–h) pair states gives rise to excess spin intermixing between hyperfine-split spin sublevels. The external magnetic field (of ultrasmall magnitude) can change the S–T intermixing rate provided by the hyperfine interaction and this way perturb the overall relative steady state populations of the spin sublevels. The authors found that the amplitude of USMFE increases with increasing temperature with the linewidth remaining unaffected. The quantitative description of temperature dependencies, however, is lacking in that study. According to our findings, the observed ultrasmall effect is rather determined by the carrier hopping process, which is in good agreement with the modeling results of the Koopmans group.40 With increasing temperature, the hopping frequency (ωhop) increases, and as a result a decrease in the MPC magnitude and an increase in the linewidth are observed.

The MPC signal, measured at 290 K and 20 mT magnetic field strength as a function of forward (negative) and reverse (positive) bias voltage, is depicted in Fig. 2c. At a bias voltage below (above) Voc, the magnetic field effect is positive (negative) and shows a weak dependence on the applied voltages while at voltages close to Voc, large values of MFE and a change of sign occur. A similar MFE on photocurrent in P3HT:PCBM bulk-heterojunction solar cells has been reported by Shakya et al.42 and Lei et al.43 A great enhancement of the magnitude of the MPC signals at bias voltages around Voc appears due to vanishing j(0) (in formula (1)) as the bias voltage approaches Voc. Moreover, we note that no mutual exciton–exciton interactions occur in the SQ layer as the photocurrents are proportional to the light intensities used in our measurements and therefore the MPC signals do not depend on the incident photon flux (see Fig. 2d).

MFEs in squaraine:fullerene bulk-heterojunction solar cells

The dependence of MFEs as a function of fullerene concentration in SQ:PC60BM bulk-heterojunction solar cells has been investigated (Fig. 3a). It can be seen that even a small amount of PC60BM (0.1 wt%) results in a decrease of a positive low-field component (BLFE = 4 mT) which indicates that PC60BM effectively quenches excitons in the SQ layer. Further, the increase in fullerene concentration (up to 10 wt%) leads to the formation of CT states (where the electron resides on PC60BM and the hole occupies the SQ molecule) which are lower in energy than (e–h) pair states as well as singlet and triplet molecular states of both SQ and PC60BM (cf.Fig. 5). The MPC signal with only the low-field component (BLFE = 4 mT), similar to the pristine SQ devices, can be easily explained by the EHP model. The sign change (from positive to negative), however, indicates that the dissociation of the electronic excited states proceeds this time via the CT states and consequently the magnetic field dependent intersystem crossing in CT states (ISCCT) plays a crucial role. Further doping the SQ with PC60BM (up to 1[thin space (1/6-em)]:[thin space (1/6-em)]6 organic to fullerene wt ratio) reduces the magnitude of the low-field (BLFE = 8 mT) negative MPC signal. In polymer:fullerene blends with such a high PC60BM concentration, phase separation usually occurs;22,30 then the corresponding MPC signal may be ascribed to the bipolaron (BP) model44 wherein HFM in charge carrier pairs with the same signs (e–e or h–h) occurs. The atomic force microscopy (AFM) analysis of the SQ:PC60BM blends indicates that fullerene is still homogeneously distributed throughout the squaraine (see Fig. S5f in the ESI). Therefore, the bipolaron model operating in the PC60BM domains may be safely neglected. We believe that here hyperfine spin mixing in CT states (EHP model) is still valid while the decrease of the MPC signal, in comparison to the small PC60BM concentration system (10 wt% PC60BM), is associated with reduction in the amount of active SQ:PC60BM interfaces.
image file: c7tc05033j-f3.tif
Fig. 3 The magnetic field effect on the photocurrent for bulk-heterojunction solar cells. The MPC signal as a function of magnetic field strength for various SQ to PC60BM weight ratios at 290 K (a); the MPC signal for SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM 1[thin space (1/6-em)]:[thin space (1/6-em)]6 wt% in the medium (0–400 mT) field range for various temperatures (b). The MPC signal for SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM 1[thin space (1/6-em)]:[thin space (1/6-em)]6 wt% versus: forward (negative) and reverse (positive) bias voltage (c), and photon flux (d) at 20 mT-magnetic field and 200 K temperature are also shown. The solid lines in part (a) and (b) represent fit according to the single- and double-Lorentzian functions. In part (e) the light intensity dependence of the photocurrent is displayed – the solid line represents a linear fit.

The half-width of magnetic field dependence of the MPC signal for samples with a high PC60BM content should be essentially lower than that in pristine SQ due to the vanishing nuclear magnetic moment of 12C as reported for the MDMO-PPV:PC60BM system.22 Although in our investigation a completely opposite behavior (BLFE = 8 mT and BLFE = 4 mT for blends and pristine SQ, respectively) is observed, this finding corroborates our previous assignment of carrier hopping frequency to the intermediate regime (ωhopωhf). Then, the MPC linewidths are determined not only by the hyperfine field but also by the lifetimes of diffusively moving e–h pairs in a solid environment exhibiting various types of structural disorders.40

In the following section, we focus on the SQ[thin space (1/6-em)]:[thin space (1/6-em)]PC60BM 1[thin space (1/6-em)]:[thin space (1/6-em)]6 wt ratio system for which the highest photoconversion efficiency has been reported so far. The negative MPC signal recorded at various temperatures (200, 250 and 290 K) saturates on the low-field scale and remains constant until about 200 mT. Above 200 mT, the absolute MPC value starts to decrease and at about 600 mT, the signal changes the sign (Fig. 3b and 4a). With increase in the magnetic field strength the positive high-field component, rather weakly affected by the temperature, increases and does not saturate up to 9 T. Moreover, upon reducing the temperature the amplitude of the low-field negative component increases and its linewidth decreases (BLFE = 8 mT and BLFE = 4 mT for 290 and 200 K, respectively) which once again emphasizes the substantial influence of the thermally activated hopping frequency on the MPC signals (Fig. 3b).

In order to clarify the origin of the high-field component we have measured the Lande g factor for a hole (gh) localized onto the SQ2 molecule by electron paramagnetic resonance, EPR (Fig. S3, ESI) and compared it with the g factor value taken from the literature for an electron (ge) placed on a PC60BM cage.45 It was found that gh = 2.0042 and ge = 1.9996, leading to Δg = 0.0046. The presence of only a positive MPC signal in the high magnetic field range and a difference in Lande factors of the order of 10−3 indicate that the so called Δg mechanism may be potentially involved as similarly considered for the MPC effect in P3HT:PC60BM photovoltaic cells (Δg = 0.002).23 In the EHP model a difference in the precession frequency (Δωp) can be induced not only by a difference in the local (of hyperfine origin) magnetic fields (ΔB) experienced by electron and hole entities forming (e–h) pairs (Δωp = BΔB/ħ) but also by a difference in the g factors (Δωp = ΔBB/ħ), where μB is the Bohr magneton, B is the applied field strength and ħ is the reduced Planck constant. Therefore, in the Δg mechanism the applied magnetic field enhances the intersystem crossing (ISCCT) between the singlet and m = 0 triplet CT states and hence the corresponding MPC signal has a sign opposite to the LFE induced by HFM.40 The characteristic lack of pronounced temperature dependence of the high-field component amplitude is a natural consequence of the spin precession – carrier hopping frequency interrelation when the external magnetic field of the order of several tesla significantly enhances the spin precession frequency, ωp. Therefore, in this regime the precession frequency is definitely greater than the charge carrier hopping frequency (ωpωhop); hence a temperature induced change in the hopping rate has no noticeable effect on the MPC signal (Fig. 4a). This outcome is significantly different from the results reported for the P3HT:PC60BM devices where the signals are strongly temperature dependent.23,46 To better rationalize the role of CT states in the dissociation/recombination decay process, the MPC data at various temperatures were fitted by the Lorentzian functions by means of formulas (2) and (3). Reasonably good fits in the high magnetic field range were obtained using the double-Lorentzian function. According to the formula

 
image file: c7tc05033j-t2.tif(4)
the CT state relaxation times for 290 K are estimated to be 0.82 ns and 0.06 ns. Nevertheless, a more appropriate approach should take into account a distribution of relaxation times. It was shown that for a non-exponential (dispersive) relaxation process the Lorentzian factor, Re[1/(1 + pτ)], may be replaced by the Cole–Cole function, Re[1/(1 + pτ)α], where Re denotes the real part, τ represents an average relaxation time and α ≤ 1 is the dispersive parameter.23,47 This can be equivalently interpreted in terms of the Kohlrausch–Williams–Watts (KWW) function, the so-called stretched exponential decay, exp(−t/τ)β with the dispersive β parameter approximately equal to the α parameter from the Cole–Cole formula (according to ref. 48αβ1.23). The respective fit depicted in Fig. 4b gives α ≈ 0.5 for temperatures 200–290 K which indicates a 3D random distribution of CT state decay times.11 The rather weak dependence of the MPC signals with temperature increase results from a gradual decrease in relaxation time (from τ = 1.88 ns for 200 K to τ = 0.57 ns for 290 K) due to the enhancement of all non-radiative decay pathways of CT states including dissociation and geminate recombination processes.


image file: c7tc05033j-f4.tif
Fig. 4 The magnetic field effect on the photocurrent for bulk-heterojunction solar cells. The MPC signals as a function of magnetic field strength for various temperatures are displayed. In part (a) the low-field negative component is approximated using a single-Lorentzian function and the high-field positive component – using a double-Lorentzian function (solid lines). For comparison, in part (b) the best fit of the high-field component applying the dispersive relaxation model is shown.

At a high magnetic field, besides the Δg mechanism, the thermal spin polarization associated with magnetic field dependence of formation probabilities of the singlet and triplet states populated according to the Boltzmann statistics may be also effective. Therefore the spin statistics of the CT states can be substantially controlled by spin-polarizing carriers using high magnetic fields and low temperatures, where the Zeeman energy is comparable with the thermal energy.49 Therefore, we can calculate the relevant polarizing parameter b = BB/kBT (where kB is the Boltzmann constant) which gives the maximum spin polarization contribution to the MPC signal as high as b2/4 ≈ 0.09% (for g = 2, B = 9 T and T = 200 K). Hence, comparing this value to the experimentally obtained MPC ≈ 2.2% at B = 9 T we can safely neglect spin polarization effects in our considerations. However, this mechanism might be of importance at cryogenic temperatures.46,49

The MPC signals, just like in the case of single-layer solar cells, do not depend on the incident photon flux and again significantly increase and change the sign at the bias voltage around Voc (Fig. 3d and c, respectively). Moreover, the photocurrent intensity is a linear function of the photon flux (Fig. 3e).

The mechanism of magnetic field effects

We consider the following mechanism for explaining the observed magnetic field effects on the photocurrent in the investigated squaraine:fullerene systems. According to the scheme based on the commonly applied EHP model, the current generation under SQ photoexcitation in single-layer solar cells (Fig. 5a) proceeds through an intermediate state of e–h pairs. The initially formed 1(e–h) pairs are quasi-degenerate in energy with 3(e–h) pairs due to the relatively weak electrostatic exchange interactions at a larger distance between electron and hole entities compared to that within the molecular (Frenkel type) exciton states. It has been established that singlet/triplet states of e–h pairs (or polaron pairs) can be mixed with each other by hyperfine interactions. When an external magnetic field greater than the hyperfine coupling strength is applied, the Zeeman splitting of triplet states removes the degeneracy between m = ±1 triplet and singlet states, and thus suppresses the 1(e–h) → 3(e–h) intersystem crossing increasing the population of 1(e–h) pairs. In single component organic solids the dissociation rates (k−1; k−3) differ due to the fact that the singlet pair state has a stronger ionic character than triplets and therefore singlet pair states are more strongly coupled with the ionic reaction products of the separated holes and electrons,50 and consequently, to explain MFEs consistently, the k−1 > k−3 relation is usually assumed.20,36 In addition, the lower-energy molecular triplet exciton states (operating here as the ‘triplet drain’19) open an efficient recombination pathway for 3(e–h) pairs, and this way limit the photocurrent. Further, for MFEs the Lorentzian shape with B1/2 width of the hyperfine scale is adequate here (cf.Fig. 2a) as shown by the quantum mechanical calculations based on Hamiltonian containing the electronic Zeeman interaction with the external magnetic field and the hyperfine interaction between a single electronic and nuclear dipole.51
image file: c7tc05033j-f5.tif
Fig. 5 Energy diagram of the relevant excited states involved in the photocurrent generation (k−1 and k−3 reaction rate constants) and charge carrier recombination (k1 and k3 reaction rate constants) processes for single-layer (a) and bulk-heterojunction (b) solar cells.

On the other hand, in bulk-heterojunction solar cells, the excited SQ singlet molecular excitons are effectively quenched into 1CT states. Intersystem crossing between singlet and triplet CT states results in the formation of long-lived 3CT states which are spin-protected from recombination to the ground state due to energetically inaccessible higher lying triplet molecular excitonic states of both squaraine and fullerene molecules (Fig. 5b). Contrary to the single component, in electron donor-electron acceptor systems with such energy level alignment the 3CT dissociation channel in the photocurrent generation process is decisive, as recently reported for similar photovoltaic systems.42 Accordingly, the long-lived triplet CT states exhibit a reduced geminate recombination rate and thus enhanced dissociation ability into free charge carriers in P3HT:PCBM devices. Moreover, a comprehensive investigation on the m-MTDATA:3TPYMB system including the direct measurements of CT state fluorescence and photocurrent generation under a magnetic field at various pressures also indicates more efficient triplet channels.19 A similar outcome has been obtained for polymer (P3HT or MDMO-PPV):fullerene blends at cryogenic temperatures where MFEs of spin polarization origin were investigated.52

D Conclusions

In summary, a study of photocurrent in SQ based solar cells, using a wide range of magnetic fields and temperatures, reveals that depending on the electron acceptor content, the photocurrent generation is limited by dissociation/recombination of e–h pairs or CT states. In a weak external magnetic field, the HFM mechanism operates, where an asymmetry in local (of hyperfine origin) magnetic fields experienced by electrons and holes forming (e–h) pairs induces spin dephasing of the magnetic dipoles. On the other hand, in a strong magnetic field the photocurrent is affected by the Δg mechanism with spin dephasing ascribed to different Lande factors of electron and hole entities. The temperature dependence of the MPC signal indicates that charge carrier hopping in a disordered environment plays the essential role in a consistent analysis of MFEs in SQ:PCBM systems. The best performance parameters of solar cells are achieved in systems without molecular ‘triplet drains’ where the dissociation of spin-protected triplet CT states is more favorable. The findings of the present work enable novel methods to engineer photovoltaic devices utilizing squaraine derivatives, in particular, to determine the alignment of energy levels involved in the photocurrent generation process.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the Polish Ministry of Science and Higher Education under “Diamond Grant” [0228/DIA/2013/42]. One of the authors (M. K.) was supported by the Foundation for Polish Science (FNP). The authors would like to acknowledge Dr Mirosław Sawczak from The Szewalski Institute of Fluid-Flow Machinery PASci for assistance with phosphorescence measurements, Dr Qihang Qin from Aalto University for assistance with AFM measurements, Dr Fredrik Pettersson from Åbo Akademi for assistance with device fabrication and Dr Tomi Elovaara from the University of Turku for assistance with the magneto-transport measurements. S. M. acknowledges financial support from the Academy of Finland [Project No. 13293916] and Jenny and Antti Wihuri Foundation. The project used experimental facilities of the Nanomicroscopy Centre (NMC) of Aalto University, Åbo Akademi University and Wihuri Physical Laboratory, University of Turku, Finland. This work was partially supported by the SuPREME project that has received funding from the European Union's Horizon 2020 research and innovation programme under Grant Agreement Number 692197.

References

  1. G. Chen, H. Sasabe, T. Igarashi, Z. Hong and J. Kido, J. Mater. Chem. A, 2015, 3, 14517–14534 CAS.
  2. T. Goh, J. S. Huang, K. G. Yager, M. Y. Sfeir, C. Y. Nam, X. Tong, L. M. Guard, P. R. Melvin, F. Antonio, B. G. Bartolome, M. L. Lee, N. Hazari and A. D. Taylor, Adv. Energy Mater., 2016, 6, 1600660 CrossRef.
  3. K. Law, J. Phys. Chem., 1987, 91, 5184–5193 CrossRef CAS.
  4. F. Silvestri, M. D. Irwin, L. Beverina, A. Facchetti, G. A. Pagani and T. J. Marks, J. Am. Chem. Soc., 2008, 130, 17640–17641 CrossRef CAS PubMed.
  5. X. Xiao, G. Wei, S. Wang, J. D. Zimmerman, C. K. Renshaw, M. E. Thompson and S. R. Forrest, Adv. Mater., 2012, 24, 1956–1960 CrossRef CAS PubMed.
  6. G. Chen, H. Sasabe, Z. Wang, X. F. Wang, Z. Hong, Y. Yang and J. Kido, Adv. Mater., 2012, 24, 2768–2773 CrossRef CAS PubMed.
  7. J. D. Zimmerman, B. E. Lassiter, X. Xiao, K. Sun, A. Dolocan, R. Gearba, D. A. Vanden Bout, K. J. Stevenson, P. Wickramasinghe, M. E. Thompson and S. R. Forrest, ACS Nano, 2013, 7, 9268–9275 CrossRef CAS PubMed.
  8. Y. Zheng and F. Wudl, J. Mater. Chem. A, 2014, 2, 48–57 CAS.
  9. G. Wei, R. R. Lunt, K. Sun, S. Wang, M. E. Thompson and S. R. Forrest, Nano Lett., 2010, 10, 3555–3559 CrossRef CAS PubMed.
  10. G. Wei, S. Wang, K. Sun, M. E. Thompson and S. R. Forrest, Adv. Energy Mater., 2011, 1, 184–187 CrossRef CAS.
  11. A. Köhler and H. Bässler, Electronic Processes in Organic Semiconductors, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2015 Search PubMed.
  12. G. D. Wei, S. Y. Wang, K. Renshaw, M. E. Thompson and S. R. Forrest, ACS Nano, 2010, 4, 1927–1934 CrossRef CAS PubMed.
  13. A. R. bin Mohd Yusoff, S. J. Lee, H. P. Kim, F. K. Shneider, W. J. da Silva and J. Jang, Adv. Funct. Mater., 2014, 24, 2240–2247 CrossRef CAS.
  14. K. Vandewal, K. Tvingstedt, A. Gadisa, O. Inganäs and J. V. Manca, Nat. Mater., 2009, 8, 904–909 CrossRef CAS PubMed.
  15. S. R. Yost, J. Lee, M. W. B. Wilson, T. Wu, D. P. McMahon, R. R. Parkhurst, N. J. Thompson, D. N. Congreve, A. Rao, K. Johnson, M. Y. Sfeir, M. G. Bawendi, T. M. Swager, R. H. Friend, M. A. Baldo and T. Van Voorhis, Nat. Chem., 2014, 6, 492–497 CrossRef CAS PubMed.
  16. D. N. Congreve, J. Lee, N. J. Thompson, E. Hontz, S. R. Yost, P. D. Reusswig, M. E. Bahlke, S. Reineke, T. Van Voorhis and M. A. Baldo, Science, 2013, 340, 334–337 CrossRef CAS PubMed.
  17. A. Rao, P. C. Y. Chow, S. Gélinas, C. W. Schlenker, C.-Z. Li, H.-L. Yip, A. K.-Y. Jen, D. S. Ginger and R. H. Friend, Nature, 2013, 500, 435–439 CrossRef CAS PubMed.
  18. S. D. Dimitrov, S. Wheeler, D. Niedzialek, B. C. Schroeder, H. Utzat, J. M. Frost, J. Yao, A. Gillett, P. S. Tuladhar, I. McCulloch, J. Nelson and J. R. Durrant, Nat. Commun., 2015, 6, 6501 CrossRef PubMed.
  19. W. Chang, D. N. Congreve, E. Hontz, M. E. Bahlke, D. P. McMahon, S. Reineke, T. C. Wu, V. Bulović, T. Van Voorhis and M. A. Baldo, Nat. Commun., 2015, 6, 6415 CrossRef PubMed.
  20. B. Hu and Y. Wu, Nat. Mater., 2007, 6, 985–991 CrossRef CAS PubMed.
  21. Z. Xu and B. Hu, Adv. Funct. Mater., 2008, 18, 2611–2617 CrossRef CAS.
  22. P. Janssen, M. Cox, S. H. W. Wouters, M. Kemerink, M. M. Wienk and B. Koopmans, Nat. Commun., 2013, 4, 2286 CAS.
  23. A. H. Devir-Wolfman, B. Khachatryan, B. R. Gautam, L. Tzabary, A. Keren, N. Tessler, Z. V. Vardeny and E. Ehrenfreund, Nat. Commun., 2014, 5, 4529 CAS.
  24. D. Sun, E. Ehrenfreund and Z. Valy Vardeny, Chem. Commun., 2014, 50, 1781–1793 RSC.
  25. C. Zhang, D. Sun, C.-X. Sheng, Y. X. Zhai, K. Mielczarek, A. Zakhidov and Z. V. Vardeny, Nat. Phys., 2015, 11, 427–434 CrossRef CAS.
  26. P. Y. Lin, T. Wu, M. Ahmadi, L. Liu, S. Haacke, T. F. Guo and B. Hu, Nano Energy, 2017, 36, 95–101 CrossRef CAS.
  27. M. A. Cabero Zabalaga, J. Wei, H. Yang, B. B. Fan, Y. Sun and W. Zhao, ACS Omega, 2017, 2, 7777–7783 CrossRef CAS.
  28. M. Klein, R. Pankiewicz, M. Zalas and W. Stampor, Sci. Rep., 2016, 6, 30077 CrossRef CAS PubMed.
  29. T. Geiger, S. Kuster, J. H. Yum, S. J. Moon, M. K. Nazeeruddin, M. Grätzel and F. Nüesch, Adv. Funct. Mater., 2009, 19, 2720–2727 CrossRef CAS.
  30. D. Veldman, I. Pek, S. C. J. Meskers, J. Rgen Sweelssen, M. M. Koetse, S. C. Veenstra, J. M. Kroon, S. S. Van Bavel, J. Loos and R. A. J. Janssen, J. Am. Chem. Soc., 2008, 130, 7721–7735 CrossRef CAS PubMed.
  31. B. Fan, Y. Maniglio, M. Simeunovic, S. Kuster, T. Geiger, R. Hany and F. Nüesch, Int. J. Photoenergy, 2009, 581068 Search PubMed.
  32. D. Pelczarski, P. Grygiel, K. Falkowski, M. Klein and W. Stampor, Org. Electron., 2015, 25, 362–376 CrossRef CAS.
  33. H. Hayashi, Introduction to Dynamic Spin Chemistry, World Scientific, 2004 Search PubMed.
  34. R. P. Groff, A. Suna, P. Avakian and R. E. Merrifield, Phys. Rev. B: Condens. Matter Mater. Phys., 1974, 9, 2655–2660 CrossRef CAS.
  35. E. Frankevich, A. Zakhidov, K. Yoshino, Y. Maruyama and K. Yakushi, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 53, 4498–4508 CrossRef CAS.
  36. J. Kalinowski, J. Szmytkowski and W. Stampor, Chem. Phys. Lett., 2003, 378, 380–387 CrossRef CAS.
  37. M. Cox, P. Janssen, F. Zhu and B. Koopmans, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 88, 1–7 CrossRef.
  38. A. Kadashchuk, V. I. Arkhipov, C. H. Kim, J. Shinar, D. W. Lee, Y. R. Hong, J. Il Jin, P. Heremans and H. Bässler, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 76, 235205 CrossRef.
  39. B. Hu, L. Yan and M. Shao, Adv. Mater., 2009, 21, 1500–1516 CrossRef CAS.
  40. A. J. Schellekens, W. Wagemans, S. P. Kersten, P. A. Bobbert and B. Koopmans, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 75204 CrossRef.
  41. T. D. Nguyen, B. R. Gautam, E. Ehrenfreund and Z. V. Vardeny, Phys. Rev. Lett., 2010, 105, 166804 CrossRef CAS PubMed.
  42. P. Shakya, P. Desai, T. Kreouzis, W. P. Gillin, S. M. Tuladhar, A. M. Ballantyne and J. Nelson, J. Phys.: Condens. Matter, 2008, 20, 452203 CrossRef.
  43. Y. Lei, Q. Song, Y. Zhang, P. Chen, R. Liu, Q. Zhang and Z. Xiong, Org. Electron., 2009, 10, 1288–1292 CrossRef CAS.
  44. P. A. Bobbert, T. D. Nguyen, F. W. A. Van Oost, B. Koopmans and M. Wohlgenannt, Phys. Rev. Lett., 2007, 99, 216801 CrossRef CAS PubMed.
  45. V. I. Krinichnyi, E. I. Yudanova and N. G. Spitsina, J. Phys. Chem. C, 2010, 114, 16756–16766 CAS.
  46. B. Khachatryan, A. H. Devir-Wolfman, L. Tzabari, N. Tessler, Z. V. Vardeny and E. Ehrenfreund, Phys. Rev. Appl., 2016, 5, 44001 CrossRef.
  47. B. Khachatryan, A. H. Devir-Wolfman, L. Tzabary, A. Keren, N. Tessler, Z. V. Vardeny and E. Ehrenfreund, Synth. Met., 2015, 60, 8–11 Search PubMed.
  48. F. Alvarez, A. Alegria and J. Colmenero, Phys. Rev. B: Condens. Matter Mater. Phys., 1991, 44, 7306–7312 CrossRef.
  49. J. Wang, A. Chepelianskii, F. Gao and N. C. Greenham, Nat. Commun., 2012, 3, 1191 CrossRef PubMed.
  50. M. Wohlgenannt and Z. V. Vardeny, J. Phys.: Condens. Matter, 2003, 15, R83–R107 CrossRef CAS.
  51. Y. Sheng, T. D. Nguyen, G. Veeraraghavan, Ö. Mermer, M. Wohlgenannt, S. Qiu and U. Scherf, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 045213 CrossRef.
  52. M. K. Etherington, J. Wang, P. C. Y. Chow and N. C. Greenham, Appl. Phys. Lett., 2014, 104, 063304 CrossRef.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c7tc05033j

This journal is © The Royal Society of Chemistry 2018