Amorphous Oxide Alloys As Interfacial Layers with Broadly Tunable Electronic Structures for Organic Photovoltaic Cells

Amorphous oxide alloys as interfacial layers with broadly tunable electronic structures for organic photovoltaic cells

Nanjia Zhoua,b,1, Myung-Gil Kimc,1, Stephen Loserc, Jeremy Smithc, Hiroyuki Yoshidad, Xugang Guoa,b, Charles Songc, Hosub Jine, Zhihua Chenf, Seok Min Yoonc, Arthur J. Freemane, Robert P. H. Changa,b,2, Antonio Facchettic,f,2, and Tobin J. Marksa,b,c,2

aDepartment of Materials Science and Engineering, Northwestern University, Evanston, IL 60208; bArgonne Northwestern Solar Energy Research Center, Northwestern University, Evanston, IL 60208; cDepartment of Chemistry and the Materials Research Center, Northwestern University, Evanston, IL 60208; dGraduate School of Advanced Integration Science, Chiba University, Chiba, 263-8522, Japan; eDepartment of Physics, Northwestern University, Evanston, IL 60208; and fMaterials Department, Polyera Corporation, Skokie, IL 60077

Contributed by Tobin J. Marks, May 9, 2015 (sent for review March 5, 2015; reviewed by David Ginger and David S. Ginley) In diverse classes of organic optoelectronic devices, controlling electron-transporting (ET) IFLs (2). These IFLs function to ac- charge injection, extraction, and blocking across organic semi- commodate the cell built-in fields, to assist carrier extraction, and conductor–inorganic electrode interfaces is crucial for enhancing to suppress parasitic carrier recombination (2, 14–16). Indeed, quantum efficiency and output voltage. To this end, the strategy studies on interfacial modification of HT electrodes verify that of inserting engineered interfacial layers (IFLs) between electrical changing the band alignment profoundly affects OPV open circuit contacts and organic semiconductors has significantly advanced voltage (Voc) and carrier lifetimes (17). Over the past decade, organic light-emitting diode and organic thin film transistor per- extensive efforts have focused on band gap engineering of OPV formance. For organic photovoltaic (OPV) devices, an electronically active layer materials, mainly by tuning the donor material highest flexible IFL design strategy to incrementally tune energy level occupied molecular orbital (HOMO) and lowest unoccupied

matching between the inorganic electrode system and the organic molecular orbital (LUMO) energetics to enhance built-in poten- CHEMISTRY photoactive components without varying the surface chemistry tials and maximize photon harvesting efficiency (18, 19). Conse- would permit OPV cells to adapt to ever-changing generations quently, compatible HT IFLs have been developed to tune their of photoactive materials. Here we report the implementation of valence band minima (VBM) or HOMO energies to match that of chemically/environmentally robust, low-temperature solution-pro- a specific organic donor. In contrast, because of historical limita- cessed amorphous transparent semiconducting oxide alloys, In-Ga-O tions in acceptor diversity, far less effort has been devoted to and Ga-Zn-Sn-O, as IFLs for inverted OPVs. Continuous variation of developing ET IFLs—to date, a few ET IFLs have been used in the IFL compositions tunes the conduction band minima over a inverted OPVs, primarily TiOx,ZnOx, several polymers, self-as- broad range, affording optimized OPV power conversion efficien- sembled monolayers, and cross-linked fullerenes (11, 14–16, 20– cies for multiple classes of organic active layer materials and estab- 24). Nevertheless, most of these materials have fixed band edge lishing clear correlations between IFL/photoactive layer energetics positions, significantly limiting their adaptability to emerging OPV and device performance. materials systems with acceptors having different LUMO energies. The recent emergence of promising nonfullerene OPV interface | amorphous oxide | photovoltaic | interfacial layers Significance olar to electrical energy conversion technologies have re- Sceived great attention as abundant and sustainable resources The development of system-independent and non–material-spe- (1–5). The diffuse nature of solar energy requires low-cost, large- cific interfacial layers (IFLs) to facilitate efficient charge collection area devices while maintaining high power conversion efficiency is of crucial importance for organic photovoltaic (OPV) cell per- (PCE) (3). As a universal design strategy, many of the emerging formance. Here we report a broadly applicable IFL design strat- thin film photovoltaic (PV) technologies such as bulk heterojunction egy using solution-processed amorphous oxide semiconductors (BHJ) organic, perovskite, quantum dot (QD), and CIGS (Cu-In- where their energetics can be tuned by varying the elemental Ga-Se) solar cells are fabricated using a trilayer architecture, where composition without varying the surface chemistry. Based on the light absorbers are sandwiched between two electrodes coated with energetic requirements of specific organic active layers, these various interfacial layers (IFLs) (6–9). Stringent requirements gov- oxides can be readily designed with dialed-in energy levels. Using ern ideal IFL materials design. Energetically, their respective band OPV solar cells as a test bed, we use a broad series of photoactive positions should match those of the photoinduced built-in potentials bulk heterojunction materials to demonstrate the effectiveness to provide energetically continuous carrier transport pathways and of these electronically tunable oxides for optimizing the perfor- to accommodate the maximum allowed output voltage. In re- mance of diverse OPV material sets. cent reports, PV performance enhancement via IFL energetic tuning has been demonstrated for very specific BHJ organic, Author contributions: N.Z., M.-G.K., R.P.H.C., A.F., and T.J.M. designed research; N.Z., QD, and perovskite cell compositions (6, 8, 10, 11). However, M.-G.K., S.L., J.S., H.Y., X.G., C.S., H.J., Z.C., S.M.Y., and A.J.F. performed research; N.Z., M.-G.K., S.L., J.S., H.Y., X.G., C.S., H.J., Z.C., S.M.Y., and A.J.F. analyzed data; and true IFL energetic tunability has not been achieved and offers a N.Z., M.-G.K., R.P.H.C., A.F., and T.J.M. wrote the paper. challenging opportunity to optimize device performance. Reviewers: D.G., University of Washington;andD.S.G.,NationalRenewableEnergy Fabricable from energetically diverse organic active layers, Laboratory. organic photovoltaics (OPVs) provide an excellent test bed for The authors declare no conflict of interest. tuning IFL energetics and are the subject of this study. The basic 1N.Z. and M.-G.K. contributed equally to this work. BHJ OPV architecture contains a mesoscopically heterogeneous 2To whom correspondence may be addressed. Email: [email protected], r-chang@ and isotropic, phase-separated donor–acceptor blend—astrategy northwestern.edu, or [email protected]. ∼ to overcome the relatively short exciton diffusion lengths ( 10 This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. nm) (2, 12, 13), sandwiched between hole-transporting (HT) and 1073/pnas.1508578112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1508578112 PNAS | June 30, 2015 | vol. 112 | no. 26 | 7897–7902 Downloaded by guest on September 28, 2021 acceptors—n-type small molecules and polymers having diverse (34); PTB7:P(NDI2OD-T2) [PTB7: poly[N,N’-bis(2-octyldodecyl)- electronic structures (1, 25–29)—illustrates the need for a de- naphthalene-1,4,5,8-bis(dicarboximide)-2,6-diyl]-alt-5,5′-(2,2’- sign strategy to create ET IFLs with broadly tunable energy bithiophene)] (35); and PDTG-TPD:PC71BM (poly-dithienogermole- levels. Furthermore, moving from oxides, to polymers, to self- thienopyrrolodione: PC71BM) (18). assembled materials incurs large, unpredictable variations in surface chemistry, compounding the challenge in energy level Results and Discussion positioning. Also, factors such as adsorbates, interfacial dipoles, IFL Design and Electronic Structure Characterization. Ideal OPV molecular orientations, and interface states confound predictable IFLs should achieve charge-selective transport with proper band IFL/organic semiconductor interfacial energy alignment (30) edge alignment and good optical transparency (14). For classical and challenge precise, continuous energy level tuning for tra- polycrystalline semiconductors such as GaAs-InAs, alloying chemi- ditional organic semiconductor or vacuum deposited oxide IFLs. cally similar constituents is known to achieve incremental band edge Thus, straightforwardly prepared ET IFL materials having broadly displacements (36–39). Olson et al. (39) achieved band alignment in tunable conduction band minima (CBM) and work functions for polycrystalline Zn1-xMgxOelectrodestoreducebandoffsetinpoly- adjusting band alignment would provide generalizable means to mer-oxide solar cells and to maximize the Voc. Band gap manipu- optimize current generation OPV performance and accommodate lation is also possible in a-Si1-xGex:H materials (40). Amorphous emerging organic donor/acceptor pairs. oxide semiconductors generally have excellent optical transparency, We report here an ET IFL design strategy using solution-pro- smooth surfaces, mechanical flexibility, and highly dispersed ns-state cessed amorphous metal oxide alloys. Continuous CBM tunability conduction bands with good electron mobility (41, 42). Furthermore, is achieved by alloying two or more electronically dissimilar ox- the excellent transport and optical properties of amorphous oxide ides, realized by precursor compositional adjustment. We show semiconductor-based transistors and transparent oxide conductors that band edge energies can be dialed-in for indium–gallium oxide argues that the low tail and midgap state densities should minimize (a-IGO) and gallium–zinc–tin oxide (a-GZTO) (Zn:Sn = 1:1) IFLs, OPV recombination sites and yield electron-selective IFLs (41, 42). providing CBM tunability over 3.5–4.6 eV. The resulting films have Additionally, the localized O 2p based valence bands, which are excellent chemical/environmental stability, conformal coating, and typically deep and poorly dispersed, should block hole injection from good electron mobilities, ideal for solar cells as verified with several typical organic donor HOMOs (43). Although compositional in- OPV classes where the IFL CBMs are systematically tuned to op- vestigations on amorphous oxides have been reported in Zn1-xBaxO, timize performance. This includes acceptors with LUMO energies Zn1-xSrxO, and Ga1-xZnxO systems (44, 45), systematic band align- from −4.1 eV to −3.5 eV. To our knowledge, these results are the ment tuning via doping such oxides has not been demonstrated. first example of band structure tuning in solution-processed amor- In this study, In2O3,ZnO,andSnO2 are chosen as host matrices phous oxide IFLs. Furthermore, the amorphous character enables because they have high electron mobilities and deep VBMs (46, 47) flexible OPV fabrication by coating amorphous oxide electrodes. and, judging from ITO (Fig. 1B), should offer CBMs which span Note that the oxide CBM energies and acceptor LUMO en- the LUMO energies of important classes of high-efficiency inverted ergies are verified here by low-energy inverse photoemission OPV active layers. Ga2O3 is used as a dopant to shift the host + spectroscopy (LEIPS) which provides excellent energy resolution matrix CBMs. It is known that bulk alloys of In2O3 SnO2 with and precision (∼0.1 eV) and negligible sample damage to or- Ga2O3 exhibit a broad range of amorphous phases, high optical ganics (31, 32) (see more in SI Appendix, Figs. S2 and S3). It will transparency, and good electron mobilities (41, 46). Unlike capital- be seen that OPV PV metrics closely track the IFL CBMs and intensive physical vapor deposition techniques (pulsed laser de- can be optimized for the five BHJ OPV material sets in Fig. 1B, position, sputtering, etc.), the fabrication of high-quality amorphous PBDT-BTI:PDI-CN2; PTB7:PC71BM [poly[[4,8-bis[(2-ethylhexyl) films is simple and achieved by mixing the metal precursor solutions oxy]-benzo[1,2-b:4,5-b’]dithiophene-2,6-diyl][3-fluoro-2-[(2-ethyl- in the desired ratios, spin-coating in ambient, then annealing the hexyl)-carbonyl]-thieno[3,4-b]thiophenediyl]]:[6,6]-phenyl-C71-butyric films at moderate temperatures. This process is compatible with acid methyl ester)] (33); PTB7:ICBA (PTB7: indene-C60 bisadduct) flexible plastic substrates. Fig. 1A shows a TEM cross-sectional image of a completed OPV with an a-IGO-50 (50% mol Ga) IFL. The nanobeam electron diffraction pattern reveals a halo, in- dicating amorphous character. Furthermore, mechanically flexible OPVs can be fabricated using high-conductivity, amorphous zinc- indium-tin oxide (a-ZITO) electrodes on plastic substrates (48) (SI Appendix,Fig.S4). These double-layer amorphous electrodes are well-matchedinproperties,yielding flexible devices with acceptable performance versus their rigid counterparts. Also, using these IFLs in inverted OPVs enhances device ambient stability. Thus, PTB7: PC71BM inverted solar cells fabricated with a-IGO (60% Ga) exhibit excellent temporal stability (SI Appendix,Fig.S19), compa- rable to those reported using ZnO IFLs (20). To probe IFL film microstructure, composition, and surface morphology as a function of composition, grazing incidence X-ray diffraction (GIXRD), X-ray photoemission spectroscopy (XPS), and atomic force microscopy (AFM) were applied to the solution-processed a-IGO and a-GZTO films. Fig. 2A shows film GIXRD data on Si/SiO2 substrates. Across the entire investigated composition range, the films are amorphous (Fig. 2A), with a continuous increase in optical band gap with Ga content, confirm- Fig. 1. Tuning bottom electrode (ITO) interfacial layer (IFL) energy levels for ing continuous evolution of the a-IGO and a-GZTO electronic inverted organic solar cells. (A) Inverted OPV and cross-sectional TEM image of an as-prepared ITO/a-IGO (50% Ga)/PTB7:PC BM/MoO /Ag solar cell. In- structures (Table 1; see details below). Except for a C 1s peak as- 71 3 signable to surface contamination, no extraneous elements are ob- sets show higher-resolution image and nanobeam electron diffraction pat- SI Appendix A C tern of a-IGO. (B) Chemical structures of the organic semiconductors used in served in the XPS ( ,Fig.S5 and ). Furthermore, the this study and energy levels of the various inverted organic solar cell in- characteristic O 1s features (Fig. 2B and SI Appendix,Fig.S5B and organic and organic components. D) show complete conversion of the precursors into the desired,

7898 | www.pnas.org/cgi/doi/10.1073/pnas.1508578112 Zhou et al. Downloaded by guest on September 28, 2021 VBM energy data. By fitting these data to a simple empirical formula, the CBM and VBM energies as a function of composition can be estimated to guide IFL energy level tuning (Eq. 1, where x = atomic ratio of Ga to total metal content). Here M is the CBM or VBM in eV below the vacuum level. The parameters b, MGa,andMmatrix arefittothedataandsummarizedinTable2:

Malloy = MGax + Mmatrixð1 − xÞ − bxð1 − xÞ. [1]

In addition to spectroscopic data, first-principles generalized gradient local density approximation calculations on crystalline Ga2O3, InGaO3, and In2O3 further support the CBM shifts with Ga doping and provide insights into their origin (SI Appendix, Figs. S8 and S9). The results connect the experimental band gap widening with increasing Ga composition and Ga 4s participa- tion in the CBM (Fig. 3C and SI Appendix, Fig. S9 D–F). Fur- thermore, the InGaO3 orbital contours reveal an electronic conduction pathway derived primarily from In 5s state overlap. Because posttransition metal oxide semiconductor conduction Fig. 2. Microstructure, oxygen environment, and morphology determination bands are principally composed of isotropic s states, similar trends for posttransition metal oxide semiconductor-based interfacial layer films. are expected for crystalline and amorphous variants (41, 46, 47). (A) Grazing incidence X-ray diffraction (GIXRD) of a-IGO and a-GZTO films as Beyond proper energy level alignment, effective OPV IFLs a function of composition. (B) Representative O1s XPS spectra of a-IGO and must have nonnegligible charge transport capacity, and this prop- a-GZTO films as a function of composition. (C) Representative AFM images of erty was quantified for the present films by field effect transistor a-IGO and a-GZTO films. (FET) and metal–semiconductor–metal diode measurements (SI Appendix,Fig.S11). Not surprisingly, FETs with high In contents < < condensed oxide lattice. The lowest binding energy peak at ∼530– (a-IGO, Ga 90%; a-GZTO, Ga 45%) have substantial electron CHEMISTRY μ ∼ −3 2 531 eV is assigned to the M-O-M lattice O, the peak at ∼531–532 mobilities ( e 10 to 8 cm /Vs) versus the OPV active layer μ ∼ −4 2 SI Appendix eV to M-OH or defect O, and the peak at ∼532–533 eV to blends ( e 10 cm /Vs; ,Fig.S11) (50). However, at μ adsorbed O species (46). Additionally, AFM data show that all of very high IFL Ga contents, e values fall, attributable to decreased s these films on glass/ITO substrates exhibit roughnesses <1.58 nm metal state overlap. Low-field vertical conductivities extracted (Fig. 2C and SI Appendix,Fig.S6). from diode structures also indicate better transport for the high In Regarding IFL energy level tunability, if the CBMs are prin- content films (up to 0.3 S/m). At high electric fields, the latter films cipally composed of unoccupied metal s-states, then doping the show deviations from Ohmic behavior, possibly because low in- heavy metal oxide matrices with light metal cations should raise trinsic carrier concentrations lead to space-charge limited con- the CBMs while concomitantly lowering the ionization potentials duction, electron trapping with Poole–Frenkelemissionintothe (36). Thus, the a-IGO and a-GZTO film series were investigated CBM, and/or increased injection barriers from the Al/CsF contact as a function of composition by UV photoemission spectroscopy (51, 52). Although the high-Ga content films have relatively low −4 2 −5 (UPS), optical band gap analysis (Fig. 3 A and B and SI Ap- electron mobilities (μe ≤ 10 cm /Vs) and conductivities (σ0 ≤ 10 pendix, Fig. S7), and LEIPS (SI Appendix, Figs. S2 and S3). The S/m), note that thin, low-conductivity oxide IFLs (≤20 nm) can still a-IGO and a-GZTO film optical band gaps (Fig. 3 A–C) undergo achieve efficient, selective charge extraction (53–55). From these smooth, monotonic increases with increasing Ga content, results, the Ga atomic content ≤ 45% for a-GZTO and between 20 whereas the a-IGO and a-GZTO film CBMs also upshift with and 80 wt % for a-IGO provides acceptable electron transport increasing Ga doping (Fig. 3D and SI Appendix, Fig. S7). The and will be evaluated with different OPV donor/acceptor sys- VBM and CBM data in Fig. 3D are derived from the UPS and tems (vide infra). optical band gap data, with the LEIPS data confirming the a-IGO CBM values (SI Appendix, Fig. S3). For a-IGO, increasing Organic–Oxide Interfacial Energy Level Alignment. The energy level the Ga content from 10 atom % Ga to 90 atom % Ga shifts the alignment at different organic/oxide interface structures was next CBM from −4.6 eV to −3.7 eV. Likewise, for a-GZTO, in- probed using combined UPS and LEIPS techniques. UPS mea- creasing the Ga content from 15 atom % Ga to 75 atom % Ga surements were performed on a-IGO films with 30, 60, and 80 atom shifts the CBM from −3.8 eV to −3.5 eV. The smaller a-GZTO % Ga (IGO-30, IGO-60, and IGO-80), stepwise coated with the CBM tunability may reflect a broader tail state distribution organic acceptors having different LUMO energies derived from versus a-IGO (49). SI Appendix, Table S3 summarizes the a-IGO LEIPS as described above [ICBA (∼−3.5 eV), PC71BM (∼−3.8 eV) and a-GZTO film band gap, CBM energy, work function, and (56), and PDI-CN2 (∼−4.1 eV) (SI Appendix,Fig.S2)]. The

Table 1. OPV performance parameters for ITO/IFL/Active Layer/MoO3/Ag cells with optimized Ga content 2 IFL Donor:acceptor (LUMO) Ga content, atom % CBM, eV VOC,V JSC, mA/cm FF, % Average (maximum) PCE, %

a-IGO PTB7:PC71BM (−3.8 eV) 60% Ga −3.93 0.734 15.8 69.7 8.06 (8.42)

a-GZTO PTB7:PC71BM (−3.8 eV) 30% Ga −3.76 0.735 15.1 69.0 7.65 (8.03) a-IGO PTB7:ICBA (−3.5 eV) 70% Ga −3.82 0.902 10.6 45.9 4.41 (4.86) PBDT-BTI: PDI-CN2 (−4.1 eV) 30% Ga −4.41 0.745 4.48 40.3 1.45 (1.53) PTB7:P(NDI2OD-T2) (−3.8 eV) 60% Ga −3.93 0.81 6.51 47.7 2.52 (2.55)

PDTG-TPD: PC71BM (−3.8 eV) 60% Ga −3.93 0.853 13.1 69.8 7.79 (7.94)

Note that the complete parameters for OPVs based on different active layers and using a-IGO/a-GZTO with varying Ga content are presented in SI Appendix, Figs. S15–S17.

Zhou et al. PNAS | June 30, 2015 | vol. 112 | no. 26 | 7899 Downloaded by guest on September 28, 2021 SI Appendix, Fig. S18 B–D, the Jscs, FFs, and PCEs of devices with both a-IGO and a-GZTO initially rise with increasing Ga content (although the conductivities of both IFLs are similar), then fall for higher Ga contents in both oxide families. The maximum performance occurs at 60–70 atom % Ga (CBM = ∼−3.93 to −3.82) and 30 atom % Ga (CBM = −3.76) for a-IGO and a-GZTO, respectively, which indicates that the CBM levels are optimally positioned relative to the PC71BM LUMO, as shown in Table 1 and SI Appendix, Fig. S18 and Table S6. The increased Jsc, FF, and PCE parameters can be attributed to the favorable alignment noted above between the ICBA LUMO and the IFL CBMs. The fact that both oxides function optimally when CBM energies match the LUMO energies of PC71BM provides an internal consistency check and strongly confirms the generality of the present IFL energetic tuning strategy for different oxides. In the second set of experiments, a-IGO composition effects were investigated for BHJ OPVs with acceptors having different LUMO energies. Regarding donors, PTB7 was used with ICBA (and PC71BM), whereas the low-lying HOMO polymer PBDT- BTI was used in combination with PDI-CN2 because it affords a large Voc. As noted above, broad IFL CBM tunability to accommodate current and emerging BHJ acceptors is a critical task Fig. 3. Tuning the band edges of amorphous oxide semiconductors via for next-generation OPVs. As shown in SI Appendix, Figs. S12– composition. (A) Plots showing band gap derivation from optical trans- S14, the combination of IGO-80/ICBA, IGO-60 or IGO-80/ mission spectra for a-IGO films as a function of Ga content. (B) Represen- PC71BM, and IGO-30 or IGO-60/PDI-CN2 achieves optimal tative UPS spectra of a-IGO films as a function of atom % Ga content. energy level alignment and minimum energy offsets (<0.2 eV) in (C) Optical band gap dependence on Ga content in a-IGO and a-GZTO films. the corresponding heterojunctions. From the data in Fig. 4 E–H (D) CBM and VBM energy dependences on atom % Ga content in a-IGO and SI Appendix a-GZTO films determined by UPS and the measured optical band gap. and , Fig. S16, the resulting OPVs using these three acceptors indeed exhibit maximum performance for the optimal IFL–acceptor pairing. For all three active layer/oxide combina- complete energy level diagrams of the organic-coated oxides are tions, PTB7:ICBA, PTB7:PC71BM, and PBDT-BTI:PDI-CN2, the shown in SI Appendix,Figs.S12–S14. Note that the vacuum level device characteristics agree well with the UPS/LEIPS-derived energy shifts are determined from the UPS cutoff energies of the secondary alignment results, showing optimal device metrics at 70–80, 60–70, electrons. The LUMO energies are determined by LEIPS for the and 30–60 atom % Ga, respectively. Besides the trend in PCE, the organic acceptors, whereas the oxide CBMs are from the VBM + variations in FF also closely track the ΔELUMO-acceptor – ECBM-oxide optical bandgaps (Fig. 2A and SI Appendix, Fig. S7) because the energy offsets described above. This finding confirms that interfacial oxide exciton binding energies are expected to be negligible. energy alignment has a direct consequence for device FF attributed Comparing different a-IGO/organic junctions in SI Appendix,Figs. to variation of interfacial charge transport properties which ulti- S12–S14, important differences are found in the energy offsets, mately affect the OPV series and shunt resistance (10, 17, 60, ΔELUMO-acceptor – ECBM-oxide, for the different a-IGO/organic junc- 61). From these results, the significance of IFL CBM tuning for tions. For ICBA, this offset decreases going from IGO-30 (∼0.7 eV) emerging OPV systems is clearly evident. and IGO-60 (∼0.7 eV) to IGO-80 (∼0.2 eV) (SI Appendix,Figs.S12– Finally, to probe the generality with other donor–acceptor pairs, S14). It is therefore expected that IGO-80 offers preferential energy IFL effects were investigated for OPVs based on three well-studied alignment for ICBA-based OPVs. The large energy offsets between high-performance OPV systems with PC71BM and P(NDI2OD- this acceptor and a-IGOs with Ga content <80 atom % will likely lead T2)astheacceptors.TheresultsforPTB7:PC71BM (described to thermionic losses for electron collection (39, 57). For PC71BM, the above), dithienogermole-thienopyrrolodione (PDTG-TPD):PC71BM, LUMO energy matches with the IGO-60 and IGO-80 CBMs rela- and PTB7:P(NDI2OD-T2) all-polymer solar cells (11, 18, 27) are tively well. Last, for PDI-CN2, the energy offset is minimized for compared in SI Appendix, Fig. S17 and Tables S8 and S9. Despite IGO-30 and IGO-60, whereas its LUMO energy is located well below large differences in chemical and structural properties, both E = ∼− the CBM of IGO-80, likely creating a charge injection barrier which PC71BM and P(NDI2OD-T2) have similar LUMO 3.8 eV impedes transport through these junctions and promotes surface (SI Appendix, Fig. S2) (56), and although absolute performance recombination (39, 58, 59) (SI Appendix,Figs.S12–S14). metrics differ, note that the metrics of all three OPV types consistently track each other, reaching performance maxima at Inverted OPV Performance. To investigate oxide IFL effects on 60 atom % Ga (CBM = −3.93 eV). OPV performance, device response was evaluated as a function of (i) a-IGO vs. a-GZTO IFL compositions with the same BHJ active layer donor–acceptor blend, (ii) IFL effects with BHJ Table 2. Fitting parameters for the empirical Eq. 1 describing acceptors having different LUMO energies, and (iii) the same the variation of CBM and VBM with Ga content in a-IGO and BHJ donor with acceptors possessing the similar LUMO ener- a-GZTO films

gies. In the first study, PTB7:PC71BM cells were fabricated using Material MGa,eV Mmatrix,eV b,eV both a-IGO and a-GZTO IFLs to define the composition effects. From the results in SI Appendix, Fig. S12, the energy offset be- a-IGO tween ICBA and IGO-80 is minimized, whereas lower atom % VBM 7.88 7.00 0.70 Ga causes unfavorable energy alignment with ICBA. SI Appen- CBM 3.50 4.73 0.60 a-GZTO dix, Fig. S18, shows the trend of Voc, short circuit currents (Jscs), VBM 7.70 7.37 0.08 fill factors (FFs), and PCEs for PTB7:PCBM cells fabricated CBM 3.20 3.87 −0.47 on a-IGO and a-GZTO with varying Ga contents. As shown in

7900 | www.pnas.org/cgi/doi/10.1073/pnas.1508578112 Zhou et al. Downloaded by guest on September 28, 2021 CHEMISTRY

Fig. 4. OPV device metric dependence of the indicated IFL compositions for inverted OPVs using various donor:acceptor combinations. (A–D) Device metric

summary for PTB7:ICBA, PTB7:PC71BM, and PBDT-BTI:PDI-CN2 OPVs using a-IGO IFLs. To illustrate the relative energy positions of the acceptor LUMOs and oxide CBMs, FF and PCE data are additionally plotted for (E and F) OPVs with PTB7:PC71BM as active layers and a-IGO or a-GZTO as IFLs and (G and H) OPVs with indicated five different active layer materials and a-IGO as IFL. Shaded regions indicate where maximal parameters are reached.

Although it is likely that factors such as IFL conductivity and IFL CBM and OPV active layer LUMO energies. By charac- mobility play some role in the OPV metric trends observed here, terizing the relevant interfacial electronic structures between the results in Fig. 4 A–D show that optimal device parameters IFLs with compositionally tuned CBM energies and organic ac- (indicated by shaded areas) are only realized for specific a-IGO ceptors having varying LUMO energies, IFL design guidelines and a-GZTO CBM energetic matches with the OPV photoactive emerge which highlight the importance of energy matching, cru- layers and that this trend has significant generality. Furthermore, cial for efficient carrier collection and minimizing recombination plotting the OPV metrics (FF and PCE) for different active layer losses. As a proof-of-concept, OPVs using a variety of energeti- combinations and IFLs directly as a function of the relative position cally different donor:acceptor combinations consistently show E – E of LUMO and CBM energies provides a compelling picture of the maximal performance when LUMO-acceptor CBM-oxide tends to importance of interfacial energy alignment. In Fig. 4 G and H,the zero. Note also that this versatile inorganic–organic interfacial junction energy offset parameter ΔELUMO-acceptor – ECBM-oxide cal- design strategy should be equally applicable to numerous other culated from independently obtained acceptor LUMO energies (by types of optoelectronic devices. LEIPS) and oxide CBM energies is used to compare device metrics Methods (FF and PCE) for all active layer/oxide combinations. As shown in Fig. 4 E–H, summarizing the aforementioned device combinations Details of thin-film characterization and TFT, vertical conductivity, and OPV using a-GZTO and a-IGO IFLs, it is evident that all independent device fabrication and measurements can be found in SI Appendix. series of OPVs, regardless of their chemical structures and ener- All reagents were purchased from Aldrich and used as received. For a-IGO, the In and Ga precursors were prepared by dissolving 195.4 mg In(NO3)3·5H2O getics, reasonably follow the same trend: FF and PCE initially and 177.5 mg Ga(NO ) ·5·5H O, respectively, in 10 mL 2-methoxyethanol, then increase with decreasing ΔE – E , reach a 3 3 2 LUMO-acceptor CBM-oxide adding 0.10 mL acetylacetone. After 10 min stirring, 57 μLof14.5MNH3 (aq) maximum when this quantity approaches 0 eV, and then fall after was added, and the solution was aged for 12 h (In) and 48 h (Ga). The In and passing the maximum, in accord with accepted OPV and organic Ga precursors were mixed with respective ratios. For a-GZTO, the Ga and Zn-Sn · light-emitting diode performance models (62–64). Note that slight precursors were prepared by dissolving 532.5 mg Ga(NO3)3 5.5H2O and 164.7 mg · mismatchescanbeattributedtovacuum level shifts at the organic– Zn(CH3COO)2 2H2O/144.2 mg SnCl2, respectively, in 5 mL 2-methoxyethanol, μ inorganic interfaces (30, 65) and the smaller range of CBM tunability then adding 154 L ethanolamine. The metal oxide precursor solutions were for a-GZTO versus a-IGO. Thus, tuning IFL CBMs to match ac- spin-coated at 3,500 rpm for 30 s (Laurell WS-650-23) on cleaned 300 nm SiO2/p + Si substrates and then annealed at their respective temperatures for ceptor LUMOs favors Ohmic contact for efficient carrier extraction 5 min. To achieve desired film thickness, spin-coating and annealing were and reduces bimolecular recombination losses. repeated (a-IGO, four layers; a-GZTO, one layer). Finally, the samples were annealed at the respective temperature for 30 min under dry air (a-IGO, Conclusions 300 °C; a-GZTO, 400 °C). This work demonstrates a broadly applicable n-type interfacial layer design strategy for inverted OPVs using robust, trans- ACKNOWLEDGMENTS. Research was supported in part by Argonne–North- parent, mechanically flexible, solution-processed amorphous western Solar Energy Research (ANSER) Center, an Energy Frontier Research Center funded by the US Department of Energy, Office of Science, Office of oxide semiconductor alloys. The continuous CBM tuning with Basic Energy Sciences, under Award DE-SC0001059 (to N.Z. and S.L.); the US composition possible can be appliedtoenergeticallymatchthe Department of Energy, Office of Science, Office of Basic Energy Sciences

Zhou et al. PNAS | June 30, 2015 | vol. 112 | no. 26 | 7901 Downloaded by guest on September 28, 2021 under Award DE-FG02-08ER46536 (to C.S., S.M.Y., A.J.F., and H.J.); Office of Center under National Science Foundation (NSF) Grant DMR-1121262 (to Naval Research Multi-University Research Initiative (ONR MURI) N00014-11- M.-G.K.). We thank the Integrated Molecular Structure Education and 1-0690 (to J.S.); Japan Science and Technology Office Presto and Japan So- Research Center (IMSERC) for characterization facilities supported by North- ciety for the Promotion of Science KAKENHI Grant 26288007 (to H.Y.); and western University, NSF under NSF CHE-0923236 and CHE-9871268 (1998), the Northwestern University Materials Research Science and Engineering Pfizer, and the State of Illinois.

1. Anthony JE (2010) Small-molecule, nonfullerene acceptors for polymer bulk hetero- 35. Yan H, et al. (2009) A high-mobility electron-transporting polymer for printed tran- junction organic photovoltaics. Chem Mater 23(3):583–590. sistors. Nature 457(7230):679–686. 2. Li G, Zhu R, Yang Y (2012) Polymer solar cells. Nat Photonics 6(3):153–161. 36. Wang ZJ, Tanaka I (2009) Conduction-band structures of wurtzite ZnO solid solutions 3. Lewis NS, Nocera DG (2006) Powering the planet: Chemical challenges in solar energy by first principles calculations. Mater Trans 50(5):1067–1070. utilization. Proc Natl Acad Sci USA 103(43):15729–15735. 37. Wei S-H, Zhang SB, Zunger A (2000) First-principles calculation of band offsets, optical 4. Lipomi DJ, Bao Z (2011) Stretchable, elastic materials and devices for solar energy bowings, and defects in CdS, CdSe, CdTe, and their alloys. J Appl Phys 87(3):1304–1311. conversion. Energy Environ Sci 4(9):3314–3328. 38. Dolginov LM, Eliseev PG, Mil’vidskiı˘ MG (1976) Multicomponent semiconductor solid 5. Jeong S, et al. (2012) Hybrid silicon nanocone-polymer solar cells. Nano Lett 12(6): solutions and their laser applications. Sov J Quantum Electron 6(7):1381–1393. 2971–2976. 39. Olson DC, et al. (2007) Band-offset engineering for enhanced open-circuit voltage in 6. Zhou H, et al. (2014) Photovoltaics. Interface engineering of highly efficient perov- polymer–oxide hybrid solar cells. Adv Funct Mater 17(2):264–269. skite solar cells. Science 345(6196):542–546. 40. Stutzmann M, Street RA, Tsai CC, Boyce JB, Ready SE (1989) Structural, optical, and spin 7. Snaith HJ (2013) Perovskites: The emergence of a new era for low-cost, high-efficiency properties of hydrogenated amorphous silicon-germanium alloys. JApplPhys66(2):569–592. solar cells. J Phys Chem Lett 4(21):3623–3630. 41. Kamiya T, Hosono H (2010) Material characteristics and applications of transparent 8. Chuang C-HM, Brown PR, Bulovic V, Bawendi MG (2014) Improved performance and stability amorphous oxide semiconductors. NPG Asia Mater 2(1):15–22. in quantum dot solar cells through band alignment engineering. Nat Mater 13(8):796–801. 42. Perkins JD, et al. (2010) Amorphous transparent conductors for PV applications. 2010 9. Repins I, et al. (2008) 19·9%-efficient ZnO/CdS/CuInGaSe2 solar cell with 81·2% fill 35th IEEE Photovoltaic Specialists Conference (PVSC) (Inst Electr Electron Eng, New factor. Prog Photovolt Res Appl 16(3):235–239. York), pp 000989–000991. 10. Chen L-M, Xu Z, Hong Z, Yang Y (2010) Interface investigation and engineering—Achieving 43. Walsh A, Da Silva JLF, Yan Y, Al-Jassim MM, Wei S-H (2009) Origin of electronic and = high performance polymer photovoltaic devices. J Mater Chem 20(13):2575–2598. optical trends in ternary In2O3(ZnO)n transparent conducting oxides (n 1,3,5): Hy- 11. He Z, et al. (2012) Enhanced power-conversion efficiency in polymer solar cells using brid density functional theory calculations. Phys Rev B 79(7):073105. an inverted device structure. Nat Photonics 6(9):591–595. 44. Pachoumi O, Li C, Vaynzof Y, Banger KK, Sirringhaus H (2013) Improved performance – 12. Bartelt JA, et al. (2013) The importance of fullerene percolation in the mixed regions and stability of inverted organic solar cells with sol gel processed, amorphous mixed of polymer–fullerene bulk heterojunction solar cells. Adv Energy Mater 3(3):364–374. metal oxide electron extraction layers comprising alkaline earth metals. Adv Energy – 13. Yu G, Gao J, Hummelen JC, Wudl F, Heeger AJ (1995) Polymer photovoltaic cells: Mater 3(11):1428 1436. Enhanced efficiencies via a network of internal donor-acceptor heterojunctions. Sci- 45. Ratcliff EL, et al. (2012) Surface composition, work function, and electrochemical – ence 270(5243):1789–1791. characteristics of gallium-doped zinc oxide. Thin Solid Films 520(17):5652 5663. 14. Steim R, Kogler FR, Brabec CJ (2010) Interface materials for organic solar cells. J Mater 46. Kim M-G, Kanatzidis MG, Facchetti A, Marks TJ (2011) Low-temperature fabrication of Chem 20(13):2499–2512. high-performance metal oxide thin-film electronics via combustion processing. Nat – 15. Zhou Y, et al. (2012) A universal method to produce low-work function electrodes for Mater 10(5):382 388. 47. Charlene C, Kai-Chen C, Evgeniy C, Jerzy K (2011) Crystalline In–Ga–Zn–O density of organic electronics. Science 336(6079):327–332. 16. Ma H, Yip H-L, Huang F, Jen AKY (2010) Interface engineering for organic electronics. states and energy band structure calculation using density function theory. Jpn J Appl Phys 50(9R):091102. Adv Funct Mater 20(9):1371–1388. 48. Zhou N, et al. (2014) Ultraflexible polymer solar cells using amorphous zinc-indium-tin 17. Knesting KM, et al. (2013) ITO interface modifiers can improve VOC in polymer solar oxide transparent electrodes. Adv Mater 26(7):1098–1104. cells and suppress surface recombination. J Phys Chem Lett 4(23):4038–4044. 49. Hosono H, Nomura K, Ogo Y, Uruga T, Kamiya T (2008) Factors controlling electron 18. Small CE, et al. (2012) High-efficiency inverted dithienogermole-thienopyrrolodione- transport properties in transparent amorphous oxide semiconductors. J Non-Cryst based polymer solar cells. Nat Photonics 6(2):115–120. Solids 354(19-25):2796–2800. 19. You J, et al. (2013) A polymer tandem solar cell with 10.6% power conversion effi- 50. Nakamura J-i, Murata K, Takahashi K (2005) Relation between carrier mobility and ciency. Nat Commun 4:1446. cell performance in bulk heterojunction solar cells consisting of soluble poly- 20. Sun Y, Seo JH, Takacs CJ, Seifter J, Heeger AJ (2011) Inverted polymer solar cells in- thiophene and fullerene derivatives. Appl Phys Lett 87(13):132105. tegrated with a low-temperature-annealed sol-gel-derived ZnO Film as an electron 51. O’Dwyer JJ (1973) The Theory of Electrical Conduction and Breakdown in Solid Di- transport layer. Adv Mater 23(14):1679–1683. electrics (Clarendon Press, Oxford). 21. Ahlswede E, Hanisch J, Powalla M (2007) Comparative study of the influence of LiF, 52. Lovejoy TC, et al. (2012) Band bending and surface defects in β-Ga2O3. Appl Phys Lett NaF, and KF on the performance of polymer bulk heterojunction solar cells. Appl Phys 100(18):181602. Lett 90(16):163504. 53. Irwin MD, et al. (2011) Structural and electrical functionality of NiO interfacial films in 22. Kim JY, et al. (2007) Efficient tandem polymer solar cells fabricated by all-solution bulk heterojunction organic solar cells. Chem Mater 23(8):2218–2226. processing. Science 317(5835):222–225. 54. Stubhan T, et al. (2013) Overcoming interface losses in organic solar cells by applying 23. Hsieh C-H, et al. (2010) Highly efficient and stable inverted polymer solar cells in- low temperature, solution processed aluminum-doped zinc oxide electron extraction tegrated with a cross-linked fullerene material as an interlayer. J Am Chem Soc layers. J Mater Chem A 1(19):6004–6009. – 132(13):4887 4893. 55. Ratcliff EL, et al. (2013) Investigating the influence of interfacial contact properties on 24. Litzov I, Brabec C (2013) Development of efficient and stable inverted bulk hetero- open circuit voltages in organic photovoltaic performance: Work function versus junction (BHJ) solar cells using different metal oxide interfaces. Materials 6(12): selectivity. Adv Energy Mater 3(5):647–656. – 5796 5820. 56. Yoshida H (2014) Low-energy inverse photoemission study on the electron affinities of 25. Holcombe TW, et al. (2011) Steric control of the donor/acceptor interface: Implications in fullerene derivatives for organic photovoltaic cells. JPhysChemC118(42):24377–24382. – organic photovoltaic charge generation. JAmChemSoc133(31):12106 12114. 57. Schulz P, et al. (2014) Interface energetics in organo-metal halide perovskite-based 26. Moore JR, et al. (2011) Polymer blend solar cells based on a high-mobility naph- photovoltaic cells. Energy Environ Sci 7(4):1377–1381. thalenediimide-based polymer acceptor: Device physics, photophysics and morphol- 58. Blumstengel S, et al. (2010) Band-offset engineering in organic/inorganic semi- – ogy. Adv Energy Mater 1(2):230 240. conductor hybrid structures. Phys Chem Chem Phys 12(37):11642–11646. 27. Zhou N, et al. (2014) Morphology-performance relationships in high-efficiency all- 59. Braun S, Salaneck WR, Fahlman M (2009) Energy-level alignment at organic/metal and polymer solar cells. Adv Energy Mater 4(3):1300785. organic/organic interfaces. Adv Mater 21(14-15):1450–1472. 28. Zang Y, et al. (2014) Integrated molecular, interfacial, and device engineering towards 60. Guerrero A, Ripolles-Sanchis T, Boix PP, Garcia-Belmonte G (2012) Series resistance in high-performance non-fullerene based organic solar cells. Adv Mater 26(32):5708–5714. organic bulk-heterojunction solar devices: Modulating carrier transport with fuller- 29. Mori D, Benten H, Okada I, Ohkita H, Ito S (2014) Highly efficient charge-carrier ene electron traps. Org Electron 13(11):2326–2332. generation and collection in polymer/polymer blend solar cells with a power con- 61. Cowan SR, et al. (2014) Chemically controlled reversible and irreversible extraction version efficiency of 5.7%. Energy Environ Sci 7(9):2939–2943. barriers via stable interface modification of zinc oxide electron collection layer in 30. Ishii H, Sugiyama K, Ito E, Seki K (1999) Energy level alignment and interfacial electronic polycarbazole-based organic solar cells. Adv Funct Mater 24(29):4671–4680. structures at organic/metal and organic/organic interfaces. Adv Mater 11(8):605–625. 62. Veinot JGC, Marks TJ (2005) Toward the ideal organic light-emitting diode. The 31. Yoshida H (2012) Near-ultraviolet inverse photoemission spectroscopy using ultra-low versatility and utility of interfacial tailoring by cross-linked siloxane interlayers. Acc energy electrons. Chem Phys Lett 539-540(29):180–185. Chem Res 38(8):632–643. 32. Yoshida H (2014) Measuring the electron affinity of organic solids: An indispensable 63. Scharber MC, et al. (2006) Design rules for donors in bulk-heterojunction solar cells— new tool for organic electronics. Anal Bioanal Chem 406(9-10):2231–2237. Towards 10% energy-conversion efficiency. Adv Mater 18(6):789–794. 33. Liang Y, et al. (2010) For the bright future-bulk heterojunction polymer solar cells 64. Servaites JD, Ratner MA, Marks TJ (2011) Organic solar cells: A new look at traditional with power conversion efficiency of 7.4%. Adv Mater 22(20):E135–E138. models. Energy Environ Sci 4(11):4410–4422. 34. He Y, Chen H-Y, Hou J, Li Y (2010) Indene-C(60) bisadduct: A new acceptor for high- 65. El-Sayed A, et al. (2013) Understanding energy-level alignment in donor-acceptor/ performance polymer solar cells. J Am Chem Soc 132(4):1377–1382. metal interfaces from core-level shifts. ACS Nano 7(8):6914–6920.

7902 | www.pnas.org/cgi/doi/10.1073/pnas.1508578112 Zhou et al. Downloaded by guest on September 28, 2021