Inverted, semitransparent small molecule photovoltaic cells

Semitransparent organic photovoltaic (OPV) cells are of interest due to their potential for fulfilling building integrated PV needs such as deployment on windows and other architectural surfaces. Moreover, semitransparent OPV cells can also be integrated into tandem and multi-junction structures to achieve a high power conversion efficiency (PCE) along with acceptable transparency for these applications. The PCE of small molecule semitransparent OPV cells based on bilayer or mixed heterojunction (HJ) structures have been relatively low since the former is limited by short exciton diffusion lengths, whereas the latter suffers from the lack of continuous paths for charge extraction.^1–8 In contrast, considerable progress has been made on the solution-processed semitransparent bulk heterojunctions OPV cells.^9,10 In this work, we demonstrate vacuum-deposited small molecule semitransparent OPVs based on both mixed and hybrid planar-mixed heterojunctions (PM-HJ) with inverted structures. Cathode contact is made on the substrate surface using a hole blocking/electron selective sol-gel ZnO layer as a cathode buffer on top of an ITO contact. The ZnO has a high electron mobility¹¹ of ∼10 cm²/V·s, 98% transmission in the visible and near infrared (NIR) spectral regions, and a work function of 4.5 eV,¹² leading to efficient electron collection and low optical loss. In addition, ZnO-based inverted structures eliminate thin but optically lossy metal layers that have been reported previously^1,4,7,13 although highly transparent metal-based electrodes have been recently demonstrated.¹⁴ It is worth noting that conventional OPV structures using wide energy gap molecules, e.g., bathophenanthroline (BPhen), as cathode buffers cannot employ symmetric ITO contacts due to the lack of electron-transporting defect states induced by the electrode deposition, whereas inverted structures enable the implementation of metal oxides, e.g., MoO₃, as the buffer for the top electrode to efficiently extract charge carriers without the concern of defect states. Meanwhile, the sol-gel ZnO cannot be employed on top of organic layers due to damage to active regions during the solution process.

The inverted semitransparent PM-HJ OPV cells exhibit PCE = 3.9% ± 0.2% under simulated AM 1.5G illumination at one sun intensity with an average transmission of $T¯$ = 51% ± 2% across the visible. This corresponds to >10% higher PCE than obtained for mixed HJ cells. This improvement is primarily due to improved charge collection efficiency and reduced series resistance in the PM-HJ architecture. We also demonstrate inverted semitransparent tandem cells incorporating two PM-HJ sub-cells that have absorption maxima in different regions of the solar spectrum. The optimal tandem cell reaches PCE = 5.3% ± 0.3% under simulated AM 1.5G illumination at one sun intensity, with $T¯$ = 31% ± 1% across the visible, with similar performance whether illuminated via either the top contact or substrate surfaces. These results show a clear tradeoff between transparency and efficiency. Single junction cells can have an attractive and selectable hue adapted to a particular application, whereas the more absorptive tandem cell is optimized for efficiency while taking on a neutral tone.

The photovoltaic cells were grown on glass substrates with pre-patterned ITO (4.2 mm × 3.5 mm patterns, sheet resistance: 15 Ω/sq). The materials used in OPV cells were obtained from commercial venders: zinc acetate dehydrate, 2-methoxyethanol and ethanolamine (Sigma-Aldrich), DBP, DTDCTB and BPhen (Lumtec), C₇₀ (SES Research), C₆₀ (MER) and MoO₃ (Alfa Aesar). The glass/ITO substrates were cleaned by successive ultrasonication in Tergitol^® (Sigma-Aldrich), deionized water, and a series of organic solvents, followed by ultraviolet ozone exposure for 10 min. The ITO surface was coated with ZnO deposited using a precursor solution prepared by dissolving 0.5 M zinc acetate dihydrate in 2-methoxyethanol with ethanolamine added as a stabilizer.¹⁵ The solution was passed through a 0.45 μm pore, polyvinylidene fluoride filter, and then spun-cast onto the substrates at 3000 rpm for 30 s. The film was then thermally annealed in ambient at 150 °C for 30 min. The substrates were transferred into a high vacuum chamber with a base pressure of 10⁻⁷torr where organic layers were deposited. The mixed organic layers were deposited at a total rate of 0.1 nm/s except for DBP:C₇₀ mixtures with a total deposition rate of 0.18 nm/s, whereas neat layers, including MoO₃, C₆₀ and C₇₀, were deposited at a rate of 0.1 nm/s. The layer thicknesses were measured using variable angle spectroscopic ellipsometry. The densities of organic materials were set at 1.1 g/cm³ for vacuum thermal evaporation. Top contacts consisting of 100 nm thick ITO with a sheet resistance of 30 Ω/sq were sputter-deposited at a base pressure of 7 × 10⁻⁸torr and a deposition rate of 0.04 nm/s through a shadow mask with an array of 11 mm² openings oriented perpendicular to the ITO contact patterns on the substrate, defining the active device area. Completed devices were directly transferred into a high-purity N₂-filled glove box with both H₂O and O₂ concentrations of <0.1 ppm. There, current density-voltage (J-V) and external quantum efficiency (EQE) measurements were performed. The light intensity of solar simulator (Asahi SPECTRA, HAL-320) was characterized by a National Renewable Energy Laboratory (NREL) traceable Si reference cell, with J_SC and PCE corrected for spectral mismatch.^16,17 The tandem cells were measured under simulated AM 1.5G illumination at one sun intensity (25 ± 1 °C, 1000 W/m², ASTM G173-03). The EQE measurements were performed using monochromated light from a 200 Hz chopped Xe-lamp without other light bias. The OPV cells were measured under illumination with and without masks with known apertures, and the device performances were identical in both cases. The measured J_SC for single junction cells were consistent with the integrated J_SC over the EQE spectra with a difference of <3%. Transmission spectra of unpatterned films were obtained using a spectrometer (Perkin-Elmer, LAMBDA 1050). Experimental errors quoted correspond to the deviation from the average values of three or more devices on the same substrates, or from unpatterned films deposited during the same growth.

Inverted semitransparent mixed HJ OPV cells were fabricated based on the donor, tetraphenyldibenzoperiflanthene (DBP), and the acceptor, C₇₀. The 30 nm thick DBP:C₇₀ (1:8 vol. ratio) blend has an average transmission of $T¯$ = 59% ± 2% between the wavelengths of λ = 400 nm and 700 nm, and appears red owing to its reduced long wavelength absorption (inset, Fig. 1(a)). The mixed HJ cells had the following structure: ITO/ZnO (30 nm)/DBP:C₇₀ (1:8 vol. ratio, thickness x = 30, 40, 50, 60, 70 nm)/MoO₃(20 nm)/ITO. The J-V and EQE characteristics are shown in Figs. 1(a) and 1(b) with device performance summarized in Table I. The OPV cell with x = 30 nm has a short circuit current density of J_SC = 4.8 ± 0.1 mA/cm², an open circuit voltage of V_OC = 0.88 ± 0.01 V, a fill factor of FF = 0.61 ± 0.01, and PCE = 2.6% ± 0.1% with $T¯$ = 59% ± 2% across the visible as shown in the left inset, Fig. 1(a). The cells with thicker photoactive layers exhibit increased EQE across the visible owing to enhanced absorption (Fig. 1(b)), thus leading to a correspondingly higher J_SC. While V_OC is independent of thickness, FF decreases with increasing x due to increased series resistance. Figure 2 shows a correlation between PCE and $T¯$ as a function of the photoactive layer thickness. The PCE and $T¯$ show opposite trends, with a maximum PCE = 3.5% ± 0.1% at x = 60 nm and $T¯$ = 47% ± 2% across the visible. PCE decreases with further increases in thickness owing to a reduction in FF.

To further understand the dependence of FF on x, the specific series resistance (R_SA) is obtained vs. active layer thickness by fitting the dark J-V characteristics to¹⁸ 

where J_s is the saturation current density in the dark, n is the ideality factor associated with the donor (acceptor) layer, k_B is the Boltzmann constant, T is the temperature, q is the elementary charge, and J_ph is the photocurrent density. Also, χ ∼ 1 is the ratio of the polaron-pair dissociation rate at the heterojunctions between donor and acceptor at V to its value at V = 0. We find that R_SA = 2.9 ± 0.1 Ω·cm² for 30 nm thick OPV cells and increases to 5.8 ± 0.1 Ω·cm² for 70 nm thick devices; a result of reduced charge collection efficiency arising from the lack of continuous paths for charge extraction (and hence FF) of thicker donor/acceptor mixed regions.

The inverted PM-HJ architecture consisting of a donor/acceptor mixture grown onto a neat acceptor layer is useful in reducing the active region series resistance by improving charge collection.^19,20 Thus, we replaced the x = 60 nm DBP:C₇₀ layer in the mixed HJ with C₇₀ (9 nm)/DBP:C₇₀ (51 nm, 1:8 vol. ratio) for the photoactive region. The neat C₇₀ layer thickness is roughly equal to its exciton diffusion length,²¹ leading to efficient exciton dissociation at the acceptor/blend interface. The C₇₀/DBP:C₇₀ film has $T¯$ = 51% ± 2% across the visible, which is > 10% higher than that of the mixed HJ. Figure 3(a) shows the J-V characteristics of both the mixed HJ and PM-HJ OPVs. The PM-HJ has J_SC = 7.5 ± 0.2 mA/cm²; almost the same as the mixed HJ. Both cells have the same V_OC = 0.89 ± 0.01 V as expected, whereas FF increases from 0.53 ± 0.01 for the mixed HJ to 0.58 ± 0.01 for the PM-HJ due to a decrease in R_SA from 5.0 ± 0.1 Ω·cm² to 3.8 ± 0.1 Ω·cm². Therefore, the PCE of the PM-HJ OPV cell is 3.9% ± 0.2%, an 11% increase compared to the mixed HJ.

To further understand the improved combination of transparency and efficiency of the PM-HJ architecture, we measured the internal quantum efficiency (IQE), i.e., the ratio of photogenerated carriers collected at the electrodes to the absorbed photons in the active region. The PM-HJ shows reduced absorption calculated using transfer matrices,^22,23 compared to the mixed HJ, particularly between the wavelengths of λ = 550 nm to 700 nm (see Fig. 3(b)). This results from a reduced amount of DBP in the photoactive region in the former structure. With a similar EQE for both architectures, the IQE of the PM-HJ is thus greater than that of the mixed HJ.

Based on the single junction cell results, we fabricated an inverted semitransparent tandem cell incorporating two PM-HJ sub-cells that absorb in different spectral regions. The sub-cells employed 2-((7-(5-(di-p-tolylamino)thiophen-2-yl)benzo[c][1,2,5]thiadiazol-4yl)methylene) malononitrile (DTDCTB):C₆₀ for absorbing in the NIR, and DBP:C₇₀ for blue-green absorption. The charge generation layer (CGL) between the sub-cells is comprised of MoO₃ (5 nm)/Ag(0.1 nm)/BPhen:C₆₀ (5 nm, 1:1 vol. ratio). The optimal tandem structure is ITO/ZnO (30 nm)/C₆₀ (5 nm)/1:1 DTDCTB:C₆₀ (60 nm)/CGL/C₇₀ (7 nm)/1:8 DBP:C₇₀ (55 nm)/ MoO₃ (20 nm)/ITO (100 nm) (see Fig. 4(a), inset). We also fabricated discrete front and back cells with the structure: ITO/ZnO (30 nm)/organic photoactive layer (front cell: C₆₀ (5 nm)/1:1 DTDCTB:C₆₀ (60 nm); back cell: C₇₀ (7 nm)/1:8 DBP:C₇₀ (55 nm))/MoO₃ (20 nm)/ITO (100 nm) for comparison.

Figure 4(a) shows J-V characteristics of discrete sub-cells and the tandem cells, with their performances summarized in Table I. The calculated optical absorption of the sub-cells is plotted in the inset of Fig. 4(a). The DTDCTB:C₆₀ and DBP:C₇₀ films appear green and red (see Fig. 4(b), inset), respectively, owing to their different absorption spectra, while the tandem film has a neutral appearance due to its broader absorption. Hence, depending on the needs of a particular application, single junction cells can be designed to have a pastel tint, whereas the more absorptive and efficient tandem cell has a neutral coloration.

Figure 4(b) also shows the EQE of the discrete and tandem cells. The EQE of tandem cells, which is a sum of EQE of discrete front and back cells, is used here to characterize the photon-harvesting efficiency of tandem cells across the solar spectrum. As shown in Fig. 4(b), the EQE of the tandem cell reaches > 50% at the wavelength of λ < 600 nm, and remains >30% at λ = 750 nm, indicating efficient photon harvesting across the visible and NIR spectral regions. The tandem cell V_OC = 1.70 ± 0.01 V, which is almost equal to the sum of two sub-cells indicating that the CGL is electrically lossless. Furthermore, J_SC = 6.2 ± 0.2 mA/cm² for the tandem is less than that of the individual sub-cells mainly due to the slight overlap of their individual absorption spectra. The tandem cell has FF = 0.51 ± 0.01, limited by that of the DTDCTB:C₆₀ PM-HJ. Overall, the optimized tandem cell exhibits PCE = 5.3% ± 0.3% under simulated AM 1.5 G illumination at one sun intensity, with $T¯$ = 31% ± 1% across the visible.

Previously, thin metal films have been employed as semitransparent cathodes in OPV cells.^1,4,7,13 These films, however, reflect and absorb a significant fraction of the incident light, which dramatically reduces the efficiency of the device when illuminated via the cathode vs. the anode. Several strategies have been developed to overcome these shortcomings, such as solution-processed Ag nanowires.^9,10 In our devices, the use of metal-free, transparent ITO for both contacts eliminates these reflections and optical losses. Top illuminated tandem cells have J_SC = 5.8 ± 0.2 mA/cm² compared to 6.2 ± 0.2 mA/cm² for bottom illumination, yielding PCE = 4.9 ± 0.3% vs. 5.3 ± 0.3%, respectively.

In conclusion, we demonstrated inverted semitransparent PM-HJ OPV cells with improved charge collection and reduced series resistance compared to an analogous mixed HJ. The optimal single junction cell achieves a PCE = 3.9 ± 0.2% with $T¯$ = 51% ± 2% across the visible. We also demonstrated an inverted semitransparent tandem cell with PCE = 5.3% ± 0.3% and $T¯$ = 31% ± 1% across the visible spectrum. These results illustrate the unique attributes of organic semiconductors to provide tinted or neutral density solar power generating coatings suitable for integration within the built environment.

The authors gratefully acknowledge the financial support in part from the SunShot Next Generation Photovoltaics program of the U.S. Department of Energy (EERE), Award No. DE-EE0005310 (X.X., experiments and analysis; K.L., analysis), and Nano Flex Power Corp. (S.R.F., analysis).