Article pubs.acs.org/JPCC
Cite This: J. Phys. Chem. C XXXX, XXX, XXX−XXX
Photodynamics in Metal-Chelating Tetraphenylazadipyrromethene Complexes: Implications for Their Potential Use as Photovoltaic Materials Regina DiScipio,†,‡ Geneviève Sauvé,*,† and Carlos E. Crespo-Hernández*,†,‡ †
Department of Chemistry and ‡Center for Chemical Dynamics, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, Ohio 44106, United States S Supporting Information *
ABSTRACT: Recent efforts to improve the photon conversion efficiency in organic photovoltaic devices have shifted toward producing and optimizing electron acceptor materials. In this contribution, the photodynamics of cobalt-, nickel-, and zinc-chelating tetraphenylazadipyrromethene (M(ADP)2) complexes are investigated as part of an evaluation of their potential use as electron acceptor in bulk-heterojunction organic photovoltaic devices using poly(3-hexylthiophene-2,5-diyl) (P3HT) as the electron donor. Steady-state and broad-band transient absorption spectroscopies are used in combination with quantum-chemical calculations in acetonitrile, chloroform, and toluene to investigate their excited-state dynamics and electronic relaxation mechanisms. Zn(ADP)2 shows the longest ground-state repopulation dynamics of the three complexes and the photovoltaic devices that incorporate Zn(ADP)2 exhibit the highest power conversion efficiency. A correlation is observed between the magnitude of the ground-state repopulation lifetime measured in solution and the power conversion efficiency measured in devices that use these complexes as electron acceptors.
1. INTRODUCTION In an organic photovoltaic device (OPV), electron donors and acceptors cooperatively (1) harvest photonic energy and (2) convert that photonic energy into excited charge pairs, which are then (3) separated and (4) collected.1,2 The efficacy of the collection process in total (a.k.a., the power conversion efficiency, PCE) is effected by the efficiency of each of these steps. Improvement of OPVs has focused significantly on optimization of the donor species, which is most commonly a conjugated polymer,3,4 although recently the focus has shifted toward improvements of the acceptor species.5−8 Fullerene derivatives are often chosen as the electron acceptor because they can convert photonic energy into charge-separated pairs with almost 100% efficiency,5 and this has resulted in PCEs of ∼12% in OPVs.6,7,9−11 However, fullerenes are imperfect. For example, they are expensive and do not have large molar absorptivity coefficients in the visible region. Hence, the brunt of the photonic harvesting is the work of the polymeric donor.7,9,12 It is reasonable to hypothesize that if an electron acceptor can contribute to photon collection in the spectral region of the solar spectrum above 600 nm while maintaining similar electrontransfer efficiency, then OPVs may generate more electricity. One prospective class of nonfullerene acceptors are metalchelating azadipyrromethene complexes (M(ADP)2, Scheme 1). M(ADP)2 complexes absorbs visible light13−16 and, paired with a polymeric donor such as poly(3-hexylthiophene-2,5-diyl) (P3HT), could be engineered to match the solar spectrum with the potential of maximizing electricity generation. Indeed, Zn(II) complexes of ADP modified with phenylethylenes have shown promising efficiencies in OPVs.15,17,18 Co(II) and Ni(II) © XXXX American Chemical Society
Scheme 1. Structure of M(ADP)2 Such That M = Co, Ni, or Zna
a
Regardless of metal center, the general structure is a distorted tetrahedral geometry such that the proximal phenyl of one ligand and the pyrrolic ring of the other ligand are within the van der Waals distance enabling π−π stacking.
complexes exhibit red-shifted absorption spectra compared with Zn(II), which can potentially improve light harvesting and PCE. Other steady-state properties that affect solar cell performance Special Issue: Prashant V. Kamat Festschrift Received: December 23, 2017 Revised: March 26, 2018
A
DOI: 10.1021/acs.jpcc.7b12657 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
ground-state geometry optimizations at the uB3LYP/LanL2DZ level of theory28 (see Tables S1 and S2 for optimized coordinates in vacuum). Running the computations for Co(ADP)2 and Ni(ADP)2 with an unrestricted functional was necessary because they have unfilled d-shell orbitals.13,29 VEEs for Zn(ADP)2 were performed at TD-PBE0/LanL2DZ level of theory,26,27 also using the PCM model in the same solvents as Co(ADP)2 and Ni(ADP)2, following ground-state geometry optimizations at the B3LYP/LanL2DZ level of theory28 (see Table S3 for optimized coordinates in vacuum). As we discuss in more details in the SI, TD-DFT calculations can significantly underestimated chargetransfer states in donor−acceptor molecules.30,31 In this work, the PBE0 functional was found to be a good compromise between accuracy and computational cost, as calculations at higher levels of theory for these molecules that have a large number of atoms (115 atoms) were shown to be impossible to compute within the computational resources available to us. To further examine the results from these TD-DFT calculations, we present VEEs calculated using the M05-2X functional32 in the SI, which has also been proposed to provide a good balance for calculating energies of charge-transfer states.33 These TD-M052X calculations support the results obtained the PBE0 functional, as discussed in the SI. 2.3. Steady-State Absorption. Steady-state absorption spectra were acquired at room temperature using a Cary 100 Bio spectrophotometer (Varian). Solvent-only samples provided the background correction. 2.4. Spectroelectrochemistry. Spectroelectrochemical (SEC) spectra were acquired with a Cary 50 Bio spectrophotometer (Varian) at room temperature under argon gas. Solutions containing 0.1 M tetrabutylammonium perchlorate (TBAP) as electrolyte and 0.2 mM Co(ADP)2, 0.3 mM Ni(ADP)2, or 0.5 mM (Zn(ADP)2, respectively, were prepared in chloroform. A nonaqueous Ag/Ag+ electrode provided the reference, while Pt mesh acted as the working electrode. A constant potential voltage of 1 V/−1 V, consistent with previous characterization of their oxidative and reductive potentials,14 was applied to the samples until complete oxidation/reduction of the complex was obtained. Thus, the absorption spectra of the singly oxidized and the singly reduced species could be acquired in conjunction with spectra of the neutral species under precisely identical conditions. The solutions showed no evidence of aggregation or degradation at any point, including during the generation of the singly oxidized or reduced species for each complexes. Different concentrations were used for each complex to mimic the concentrations used during the transient absorption experiments (see below), where an identical OD at the excitation wavelength of 690 nm was desired for the three complexes. Background correction was done using a 0.1 M solution of TBAP in chloroform. 2.5. Transient Absorption Spectroscopy. The transient absorption setup has been previously described in detail.34−36 A fraction of the fundamental beam from a Ti-sapphire regenerative amplified laser system (Libra-HE, Coherent; 800 nm, 100 fs, 4.0 W at 1 kHz) was converted to 690 nm using an optical parametric amplification (TOPAS, Quantronix/Light Conversion) to generate the excitation pulse. The polarization of the pump beam was randomized with a depolarizing lens and attenuated to 1 mJ at the sample with a neutral density filter. This prevents rotational effects and multiphoton signals, respectively, from contributing to the observed kinetics. The probe pulse was generated by focusing a fraction of the residual (ca. 2%) of the 800 nm fundamental beam in a 2 mm CaF2 plate translating
are similar for the three complexes: Co(ADP)2, Ni(ADP)2, and Zn(ADP)2 all have the same distorted tetrahedral geometry and show similar oxidative/reductive characteristics and electron affinities as determined computationally.13,14,16,17,19,20 Films from blends with P3HT are expected to have identical morphologies because the only change between the three acceptors is the metal center, which is surrounded by the ligands and does not interact with the polymer. We would therefore expect all M(ADP)2 complexes to perform well in OPVs, with the Co(II) and Ni(II) having slightly better performance than the Zn(II) complex. However, the metal center may strongly affect excited-state dynamics, which may strongly impact chargetransfer rates and OPV performance. For example, charge separation must occur before recombination. Charge separation has been observed within 100 ps in devices,21,22 requiring a charge recombination lifetime of no less than 100 ps and preferably on the order of nanoseconds for charge separation to be competitive.23 To our knowledge, the excited-state dynamics of these complexes are not known. In this contribution, the PCE in OPV devices using M(ADP)2 complexes with Co(II), Ni(II), and Zn(II) metal centers as electron acceptors is evaluated. Furthermore, the excited-state relaxation mechanisms of these metal complexes are investigated in three different solvents by using steady-state absorption spectroscopy, time-independent and time-dependent density functional calculations, and transient absorption spectroscopy with femtosecond time resolution. We determined that there are significant differences between the excited-state dynamics of the complexes, which were not indicated by ground-state characterization methods. Moreover, these dynamics correlate with observations of PCE in OPV devices.
2. EXPERIMENTAL AND COMPUTATIONAL METHODS 2.1. Synthesis. The azadipyrromethene-based complexes were synthesized following the published procedures.13,15,24 Results characterizing the complexes are summarized below, and the 1H NMR and MALDI-TOF spectra are shown in the Supporting Information (SI). Co(ADP)2. 1H NMR (CDCl3, 500 MHz) δ: 65.72 (bs, 4H), 16.36 (bs, 8H), 14.60 (bs, 8H), 8.92 (bs, 8H), 6.17 (bs, 4H), 4.53 (bs, 4H), −14.63 (bs, 8H) (Figure S1). MALDI-TOF MS: m/z calculated for C64H44N6Co 955.30 and experimentally determined to be 955.27 (Figure S2). Ni(ADP)2. 1H NMR (CDCl3, 500 MHz) δ: 62.92 (s, 4H), 35.37 (bs, 8H), 8.70 (bs, 8H), 7.38 (t, 8H, J = 7.02 Hz), 7.18 (bs, 4H), 0.22 (t, 4H, J = 6.90 Hz), −1.43 (d, 8H, J = 7.38 Hz) (Figure S3). MALDI-TOF MS: m/z calculated for C64H44N6Ni 954.30 and experimentally determined to be 954.27 (Figure S4). Zn(ADP)2. 1H NMR (CDCl3, 500 MHz) δ: 7.85 (dd, 8H), 7.48 (dd, 8H), 7.37−7.43 (m, 12H), 7.06−7.12 (m, 12H), 6.71 (s, 4H) (Figure S5). MALDI-TOF MS: m/z calculated for C64H44N6Zn to be 960.29 and experimentally determined to be 960.24 (Figure S6). Poly(3-hexylthiophene-2,5-diyl) (P3HT). P3HT was synthesized using the literature procedure.25 The molecular weight was estimated by gel chromatography (GPC) in tetrahydrofuran using poly(styrene) standards: Mn = 17 600 g/mol and PDI = 1.36. The degree of polymerization was ∼69, as estimated by 1H NMR end-group analysis.25 2.2. Quantum-Chemical Calculations. Vertical excitation energies (VEEs) for Co(ADP)2 and Ni(ADP)2 were performed at the TD-uPBE0/LanL2DZ level of theory26,27 using the PCM solvent model for toluene, chloroform, and acetonitrile following B
DOI: 10.1021/acs.jpcc.7b12657 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
Figure 1. Normalized ground-state absorption spectra for Co(ADP)2 (left), Ni(ADP)2 (middle), and Zn(ADP)2 (right) showing a slight blue-shifting of absorption maximum (ca. 600 nm) and a decrease in the relative intensity of the lowest-energy absorption shoulder as the polarity of the solvent increases.
Table 1. Vertical Excitation Energies for the M(ADP)2 Complexes at the Restricted (Zn(ADP)2) or Unrestricted (Co(ADP)2 and Ni(ADP)2) TD-PBE0/PCM/LanL2DZ//B3LYP/LanL2DZ Level of Theory in Acetonitrile, Chloroform, and Toluene acetonitrile Co(ADP)2
Ni(ADP)2
Zn(ADP)2
chloroform
toluene
energy (eV)
energy (nm)
oscillator strength
energy (eV)
energy (nm)
oscillator strength
energy (eV)
energy (nm)
1.96 1.99 2.13 2.14 2.17 2.21 2.24 energy (eV) 1.94 2.00 2.10 2.14 2.17 2.18 2.23 energy (eV)
632 623 582 579 571 561 554 energy (nm) 639 620 590 579 571 569 556 energy (nm)
0.005 1.96 0.006 2.00 0.002 2.14 0.001 2.14 0.112 2.17 0.530 2.20 0.853 2.22 oscillator strength energy (eV) 0.003 1.93 0.005 2.01 0.001 2.10 0.001 2.15 0.474 2.15 0.070 2.18 0.834 2.21 oscillator strength energy (eV)
632 620 579 579 571 564 558 energy (nm) 642 617 590 577 577 569 561 energy (nm)
0.006 0.007 0.006 0.007 0.275 0.388 0.886 oscillator strength 0.004 0.007 0.002 0.296 0.236 0.050 0.851 oscillator strength
1.96 2.00 2.14 2.15 2.17 2.20 2.22 energy (eV) 1.93 2.01 2.10 2.14 2.16 2.18 2.20 energy (eV)
632 0.006 620 0.008 579 0.010 576 0.020 571 0.374 564 0.292 558 0.894 energy (nm) oscillator strength 642 0.004 617 0.008 590 0.004 579 0.538 574 0.006 569 0.068 564 0.842 energy (nm) oscillator strength
2.00 2.01 2.26 2.30
620 617 549 539
620 617 554 544
0.009 0.008 0.741 0.909
2.00 2.02 2.23 2.27
0.008 0.006 0.712 0.871
2.00 2.01 2.24 2.28
620 614 556 546
oscillator strength
0.009 0.009 0.753 0.925
a Ground-state multiplicity is a quartet for Co(ADP)2, a triplet for Ni(ADP)2, and a singlet for Zn(ADP)2, with presumed equal multiplicity for the optically accessible excited states of each complex.13
perpendicular to the beam, as previously described,34,37,38 which generates a white-light continuum from ca. 320 to 700 nm. All transient absorption experiments were performed on solutions of 0.2 OD at the excitation wavelength of 690 nm. No evidence of dimerization or photodegradation was observed before or during the experiments, as judged from the absorption spectra taken before, during, and after the transient absorption experiments. Data were acquired using a homemade program using LabView (National Instruments) that has been previously described.34,35 Data analysis of the transient absorption data was performed in Igor Pro 6.36 (WaveMetrics)34,35,39 using two or three exponential functions for Zn(ADP)2 or Co(ADP)2 and Ni(ADP)2, respectively. The lifetimes were fitted globally in the spectral probe window from 400 to 630 nm. Time zero was defined at the maximum amplitude of the ground-state depopulation. 2.6. Photovoltaic Fabrication. ITO-coated glass substrates (R = 15 Ω/sq) were cleaned stepwise in soapy water, DI water,
acetone, and isopropanol under ultrasonication for 15 min, followed by a UV-ozone treatment at 80 °C for 15 min. From a ZnO precursor solution of 0.25 M zinc acetate dihydrate in 0.25 M ethanolamine and 2-methoxyethanol, a ZnO layer was prepared by spin coating at 4000 rpm for 40 s, then heat-treated at 150 °C for 7 min. The photoactive layers were spin-coated inside the glovebox (PurelabHE) at 1000 rpm for 40 s followed, by 2000 rpm for 2 s from a blend solution with a total concentration of 20 mg/mL in o-dichlorobenzene. All acceptors were blended with P3HT in a 1:1 ratio. The photoactive layers were preannealed at 120 °C for 30 min, followed by deposition of MoO3 (10 nm) and Ag (80 nm) under a vacuum pressure of 3 × 10−5 Torr using the Angstrom Engineering Evovac Deposition System. Under AM 1.5 solar illumination at an intensity of 100 mW/cm2, solar cell measurements were performed using an Oriel Sol2A solar simulator and a Keithley 2400 source meter inside the glovebox. The devices have a total effective area of 0.20 cm2. C
DOI: 10.1021/acs.jpcc.7b12657 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
Figure 2. Primary Kohn−Sham orbital transitions involved in the lowest-lying three VEEs reported in Table 1 for Co(ADP)2 in chloroform and obtained at the TD-uPBE0/PCM/LanL2DZ//B3LPY/LanL2DZ level of theory.
Figure 3. Transient absorption spectra for Co(ADP)2 (left), Ni(ADP)2 (middle), and Zn(ADP)2 (right) upon excitation at 690 nm in chloroform. The top panel shows the time delays associated with the first and second lifetimes, whereas the bottom panel shows those associated with third lifetime for each complex.
3. RESULTS AND DISCUSSION
ground-state absorption spectra for all three complexes and may be composed of more than one electronic transitions (see below), it is unclear whether the energy of this absorption shoulder change with solvent polarity. Overall, the solvatochromic effects in the spectral region from 500 to 800 nm are insignificant in going from toluene to chloroform, but the energy and intensity of the absorption bands noticeably change when the complexes are dissolved in acetonitrile, particularly for the Zn(ADP)2 complex (see Table 1). 3.2. Computations. Following ground-state geometry optimizations, vertical excitation energies (VEEs) were calculated for Co(ADP)2, Ni(ADP)2, and Zn(ADP)2 complexes in
3.1. Ground-State Solvatochromism. Co(ADP)2, Ni(ADP)2, and Zn(ADP)2 were dissolved in toluene, chloroform, or acetonitrile, and their absorption spectra were collected in each solvent. These solvents were chosen because they provide a wide range of polarities without sacrificing the solubility of the complexes. As can be seen in Figure 1, as the polarity of the solvent increases, the energy of the dominant absorption band at ca. 600 nm increases, whereas the relative intensity of the lowestenergy absorption band at ca. 670 nm decreases. Because the lowest-energy absorption band appears as a shoulder in the D
DOI: 10.1021/acs.jpcc.7b12657 J. Phys. Chem. C XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry C
Figure 4. Representative kinetic traces and best global fits of the transient absorption data for Co(ADP)2, Ni(ADP)2, and Zn(ADP)2 in acetonitrile, chloroform, and toluene, respectively. A sum of three exponentials function was used for globally fitting the Co(ADP)2 and Ni(ADP)2 broad-band data, whereas a sum of two exponentials function was used to fit the Zn(ADP)2 data.
Table 2. Global Fit Lifetimes for Co(ADP)2, Ni(ADP)2, and Zn(ADP)2 in Acetonitrile, Chloroform, and Toluenea complex
solvent
A1 (%)b
τ1 (ps)
A2 (%)b
τ2 (ps)
A3 (%)b
τ3 (ps)
Co(ADP)2 Ni(ADP)2 Zn(ADP)2
all all acetonitrile chloroform toluene
95 ± 2 74 ± 2
0.4 ± 0.1 0.5 ± 0.1
