Zwitterion Nondetergent Sulfobetaine-Modified SnO2 as an Efficient Electron Transport Layer for Inverted Organic Solar Cells.

Tin oxide (SnO2) has been widely accepted as an effective electron transport layer (ETL) for optoelectronic devices because of its outstanding electro-optical properties such as its suitable band energy levels, high electron mobility, and high transparency. Here, we report a simple but effective interfacial engineering strategy to achieve highly efficient and stable inverted organic solar cells (iOSCs) via a low-temperature solution process and an SnO2 ETL modified by zwitterion nondetergent sulfobetaine 3-(4-tert-butyl-1-pyridinio)-1-propanesulfonate (NDSB-256-4T). We found that NDSB-256-4T helps reduce the work function of SnO2, resulting in more efficient electron extraction and transport to the cathode of iOSCs. NDSB-256-4T also passivates the defects in SnO2, which serves as recombination centers that greatly reduce the device performance of iOSCs. In addition, NDSB-256-4T provides the better interfacial contact between SnO2 and the active layer. Thus, a higher power conversion efficiency (PCE) and longer device stability of iOSCs are expected for a combination of SnO2 and NDSB-256-4T than for devices based on SnO2 only. With these enhanced interfacial properties, P3HT:PC60BM-based iOSCs using SnO2/NDSB-256-4T (0.2 mg/mL) as an ETL showed both a higher average PCE of 3.72%, which is 33% higher than devices using SnO2 only (2.79%) and excellent device stability (over 90% of the initial PCE remained after storing 5 weeks in ambient air without encapsulation). In an extended application of the PTB7-Th:PC70BM systems, we achieved an impressive average PCE of 8.22% with SnO2/NDSB-256-4T (0.2 mg/mL) as the ETL, while devices based on SnO2 exhibited an average PCE of only 4.45%. Thus, the use of zwitterion to modify SnO2 ETL is a promising way to obtain both highly efficient and stable iOSCs.

Introduction

In the field of renewable energy field, bulk heterojunction solar cells (OSCs), which are normally based on conjugated polymer donors and fullerene derivative acceptors, have attracted particular attention from researchers as third generation solar cells because of their outstanding advantages (low cost abundant materials, light weight, simple solution preparation processing, and high compatibility with roll-to-roll manufacturing techniques on flexible substrates).[1−5] Considerable progress has been achieved for OSCs because of the great efforts in developing novel donor–acceptor materials and interfacial engineering.[6,7] To date, the highest power conversion efficiencies (PCEs) of OSCs with single-junction and tandem structures are approaching 16.5 and 17.3%, respectively.[8−10] However, among the three essential characteristics of photovoltaic technology (efficiency, scalability, and stability),[11] the stability of OSC devices with an average lifetime of more than 10 years is still considered as the real hurdle that OSCs need to overcome before entering the photovoltaic market.[12]

Normally, OSCs are built in two types, conventional and inverted structures.[13] Inverted organic solar cells (iOSCs) have a typical device structure consisting of five components: a cathode electrode (indium tin oxide, ITO), electron transport layers (ETLs), active layer, hole transport layers (HTLs), and anode electrode (Ag or Au metal). These architectures are more stable than conventional structures because of the self-protection from both the cathode and anode sides.[14] In iOSCs, the ETL plays a crucial role in extracting and transporting photogenerated electrons from the organic absorber layer to the ITO cathode. The ETL also functions as a hole-blocking layer and has a large effect on the overall device performance of iOSCs.[15] ETLs in high-performance iOSCs will ideally form a smooth and compact film on the ITO, have high transparency, have a suitable conduction band, and provide high electron mobility.[16]

In the past, several kinds of materials such as organic compounds,[17] inorganic compounds,[18] metallic salts,[19] and others[20] have been used as ETLs for enhancing the device performance of iOSCs. Among these materials, n-type metal oxides such as TiO2 and ZnO are the two most commonly exploited ETLs because of their high transmittance of visible light and appropriate energy band levels.[21,22] Unfortunately, the TiO2 ETL suffers from high electron recombination rates because of its relatively low intrinsic electron mobility (10–5 cm2 V–1 s–1).[23] Meanwhile, the ZnO ETL is chemically unstable and sensitive to weak acids. Both TiO2 and ZnO are unstable under UV light exposure because of their high photocatalytic activity.[24] Typically, TiO2 requires a high temperature (up to 500 °C) and long (2–5 h) annealing procedure to convert it into the conductive phase, hindering its use in final iOSC products, where roll-to-roll industrial processing on flexible substrates (temperature processing below 200 °C) is employed.[25] Furthermore, iOSCs using TiO2 or ZnO ETLs usually suffer from light-soaking problems; that is, very poor device performance of these iOSCs is observed in the absence of UV sources.[26,27]

Recently, tin oxide (SnO2) has become one of the most promising candidates for ETLs because of its outstanding properties such as high transparency, good antireflective properties, suitable conduction band, and deep valence band level (especially when compared with the TiO2). SnO2 possesses a large electron mobility of up to 240 cm2 V–1 s–1[28] and a wider band gap (3.8 eV),[29] which are favorable features for improving device performance of iOSCs. Unlike TiO2 or ZnO, iOSC devices using SnO2 were free of light-soaking and showed excellent device stability.[30,31] SnO2 has been intensively studied as one of the most promising ETLs for the eventual commercialization of perovskite solar cells (PSCs).[32,33] Although high-efficiency PSCs can be achieved with single SnO2 as an ETL, several works have also focused on improving the device performance of PSCs by modifying SnO2 with other organic materials such as PCBM.[34] Composites of SnO2 with other metal oxides have been made into ETL bilayers such as MgO,[35] ZnO,[36] and TiO2.[37] Other strategies have also been employed like doping of SnO2 with dopants like Li,[38] Sb,[39] Nb,[40] and Y,[41] which greatly improved device performance of PSCs in comparison to the pristine SnO2. Zwitterionic compound modification[42] or the use of fullerene derivatives anchored to SnO2 have also shown superior performance of PSCs.[25] Moreover, it has been reported that SnO2 can also be effectively modified by chemical bath deposition technique to achieve stable and high performance of PSCs.[43] Interestingly, in another aspect, SnO2 can be employed for interfacial modifications of TiO2 to greatly improve the electron extraction and transportation processes in PSCs, thus highlighting the wide use of SnO2 in applications for PSCs.[44]

Like PSCs, interfacial engineering of ETLs such as ZnO or TiO2 has also drawn considerable attention from researchers, seeking to achieve highly efficient and stable iOSCs.[45,46] Interfacial modifier layers or buffer layers like organic or inorganic materials are often inserted between the ETL and the active layer to facilitate electron extraction and transportation by means of reducing the work function (WF) of the ETL and decreasing the charge recombination rates in iOSCs.[20] However, to our knowledge, works on modifying SnO2 or composite SnO2 with other materials in ETL applications for iOSCs are very limited. We note here that modifying SnO2 or composites of SnO2 and other materials can work well in PSC systems, but this might not be transferable to iOSCs because of the differences in material properties between organic and perovskite (inorganic) solar cells. This may be the reason behind the limited reports on modifying SnO2 or composites of SnO2 with other materials for applications in iOSCs. Therefore, modifying SnO2 or forming composites of SnO2 with other materials in iOSC applications is of interest.

In our previous reports, we found that ionic liquids such as 1-benzyl-3-methylimidazolium chloride ([BzMIM]Cl)[47] or alkali carbonates (Li2CO3, K2CO3, and Rb2CO3)[48] can be used as interfacial modifiers for SnO2 in a facile low-temperature solution process to provide greatly improved PCEs. Shen et al. reported that a poly (9,9-n-dihexyl-2,7-fluorene-alt-9-phenyl-3,6-carbazole) (PFN) can also be used to modify SnO2 to obtain great improvements in device performances for iOSCs.[49] Other strategies such as doping of SnO2 with metals (such as Mg) has also been pursued by Huang et al. In their report, Mg-doped SnO2 was obtained by adding magnesium chloride into SnO2 precursor solutions.[50] Huang et al. demonstrated that the Mg:SnO2 interfacial layers showed great improvements in the electron extraction and reduced photogenerated carrier recombination rates for iOSCs compared to the undoped SnO2. Apart from these potential interfacial modifier materials for SnO2, we report here that zwitterion nondetergent sulfobetaine like 3-(4-tert-butyl-1-pyridinio)-1-propanesulfonate (NDSB-256-4T) can also be exploited as a very promising material to modify SnO2 to achieve highly efficient and stable iOSCs. Nondetergent sulfobetaine (NDSB-256-4T) can simply be prepared by dissolving it in methanol followed by spin-coating the solution onto SnO2 ETLs. Noticeably, NDSB-256-4T can play several important roles in the modification of SnO2 ETL: (1) it reduces the WF of SnO2, which results in better energy band alignment between SnO2 and the active layer to allow for more efficient electron extraction and transportation to the cathode; (2) it suppresses charge recombination rates in iOSCs by effectively passivating oxygen vacancy-related defects in SnO2 that occur during low-temperature solution synthesis processes, thus significantly enhancing device performance for iOSCs; (3) it pulls more electrons from the active layer to the ETL/active interface, upgrading the electron transport capacity; and (4) it builds a better interfacial contact between SnO2 and the active layer, reducing the electron transport/transfer resistance and thus greatly improving the efficiency and stability of iOSC devices. P3HT:PC60BM-based iOSCs using SnO2/NDSB-256-4T (0.2 mg/mL) as an ETL achieved both a higher average PCE of 3.72%, which is 33% higher than devices using SnO2 only (2.79%) and revealed excellent device stability. More importantly, we also obtained an impressive average PCE of 8.22% with SnO2/NDSB-256-4T (0.2 mg/mL) as the ETL in a further application of the PTB7-Th:PC70BM systems, while the devices based on SnO2 exhibited an average PCE of just 4.45%. At this point, we believe that the use of NDSB-256-4T to modify SnO2 ETLs can serve as a good strategy for building commercial high-performance and highly stable iOSCs.

Results and Discussion

The optical properties of SnO2 and SnO2/NDSB-256-4T ETLs play an important role in the device performance of iOSCs. Therefore, we first investigated the transmittance and UV-absorption for our ETL samples coated on glass substrates. Figure a shows the transmittance spectra of SnO2 and SnO2/NDSB-256-4T ETL samples, while their UV–vis absorption is presented in Figure S1 (Supporting Information). Ideally, the transmittance of ETL should be over 85% in the visible region (380–780 nm) to guarantee that the maximum sunlight is absorbed by the active layer.[51] It appears that both SnO2 and SnO2/NDSB-256-4T ETL samples exhibited an average transmittance of over 94% in the visible region (380–780 nm), highlighting the outstanding optical properties of our ETLs. Compared to the SnO2, the SnO2/NDSB-256-4T ETL sample showed a slightly enhanced transmittance because of antireflection. It has been noted that the interference from the glass substrate can be eliminated because we have used the bare glass substrate as a reference during baseline measurement. As presented in Figure a, it is clear that the transmittance of the bare glass substrate is ca. 100% in the visible region (380–780 nm), indicating that there is no interference from the glass substrate in our transmittance measurements for SnO2 and SnO2/NDSB-256-4T ETL samples. Figure b demonstrates the Tauc plots of SnO2 and SnO2/NDSB-256-4T ETL samples, as extracted from their UV–vis absorption on glass substrates (Figure S1). As displayed in Figure b, the optical band gaps of the SnO2 and SnO2/NDSB-256-4T ETL samples were found to be 3.80 and 3.84 eV, respectively.

Figure 1

(a) Transmittance, (b) (αhν)2 vs photon energy, (c) room-temperature PL studies, and (d) XRD of SnO2 and SnO2/NDSB-256-4T films deposited on glass substrates.

The photoluminescence (PL) spectroscopy measurements of SnO2 and SnO2/NDSB-256-4T ETL films coated on glass substrates are shown in Figure c using a wavelength excitation of 350 nm. Both the optical band gaps of SnO2 (3.80 eV) and SnO2/NDSB-256-4T (3.84 eV) were larger than the excitation energy (3.54 eV). Thus, any PL emission peaks were attributed to defects in our samples. As can be seen from Figure c, there are three clear emission peaks for the SnO2 and SnO2/NDSB-256-4T samples. All three clear emission peaks likely stemmed from oxygen vacancy-related defects in the SnO2 and SnO2/NDSB-256-4T samples.[52] Note that a lower PL intensity means fewer defects, and fewer defects in the ETL mean a lower charge recombination rate for iOSC devices.[53] Compared to SnO2, the SnO2/NDSB-256-4T sample showed strong PL quenching over a wide region (300–900 nm), suggesting that it effectively passivated defects in SnO2 with NDSB-256-4T.

We continued analyzing the structural properties of SnO2 and SnO2/NDSB-256-4T ETL samples via X-ray diffraction (XRD) measurements. The XRD spectra of the SnO2 and SnO2/NDSB-256-4T samples are presented in Figure d. In comparison with JCPDS card no. 41-1445, it is clear that no principal peaks among the SnO2 and SnO2/NDSB-256-4T samples can be detected, implying that our ETL samples had amorphous structure states because of low-temperature annealing, as reported by other groups.[30,31,54]

To understand the chemical states of our SnO2 and SnO2/NDSB-256-4T ETL samples, we further investigated our ETL samples using X-ray photoelectron spectroscopy (XPS). Figure a presents the XPS surveys of our ETL samples, while XPS spectra for the Sn 3d and O 1s regions are given in Figure b,c, respectively. It is clear from Figure a that the XPS survey profile of the SnO2/NDSB-256-4T ETL sample contains S, C, and N peaks, suggesting the existence of NDSB-256-4T (C12H19NO3S) on the SnO2. Figure b indicates the presence of Sn 3d5/2 and Sn 3d3/2 in the SnO2 sample, as represented by the dominant peaks at 486.92 and 495.34 eV, respectively.[55] Compared to SnO2, there are slight shifts in the Sn 3d5/2 and Sn 3d3/2 binding energies at 486.17 and 494.58 eV, respectively, for the SnO2/NDSB-256-4T samples, suggesting that there might be a chemical bond between the NDSB-256-4T and the SnO2.[55] Because of the presence of O– of the electronegative anions (SO32–) of NDSB-256-4T (C12H19NO3S) on the SnO2 surface, the O–Sn–O stretch, Sn–O vibration, and O–O stretching vibration in the SnO2 might occur, resulting in the peak shift in our XPS results as well documented in several previous reports.[42,55,56]Figure c indicates a small shift in the O 1s dominant peak binding energy of SnO2/NDSB-256-4T (530.08 eV) to a lower binding energy compared to SnO2 (530.68 eV). In the SnO2 only ETL, the atomic percentages of Sn and O were 36.85 and 63.15%, respectively. Meanwhile, the SnO2/NDSB-256-4T sample had atomic percentages of 19.45, 40.12, 34.71, 3.17, and 2.55% for Sn, O, C, N, and S, respectively. The XPS results reconfirmed significant amounts of C, N, and S elements in NDSB-256-4T (C12H19NO3S) on SnO2. Notably, a large reduction in the atomic percentage of O (40.12%) was observed for SnO2/NDSB-256-4T compared to the SnO2 ETL only (63.15%), suggesting successful passivation of oxygen vacancy-related defects in SnO2 with the NDSB-256-4T modifier. Figure d–f demonstrates the coexistences of C 1s, N 1s, and S 2p in the SnO2/NDSB-256-4T ETL sample. Figure e clearly indicates the presence of two distinguishable peaks related to the nitrogen atoms. One peak (N1) was located at a binding energy of 401.64 eV, which typically originates from protonated nitrogen (NH3+).[57] Another peak (N2) at 399.69 eV is ascribed to nonprotonated nitrogen (amide groups).[57] The XPS results have again confirmed the surface composition as well as chemical states for our ETL samples.

Figure 2

(a) XPS surveys, (b) Sn 3d, (c) O 1s of SnO2, and SnO2/NDSB-256-4T ETL samples. (d) C 1s, (e) N 1s, and (f) S 2p of SnO2/NDSB-256-4T samples.

Next, we studied the morphology of bare ITO, SnO2, and SnO2/NDSB-256-4T ETLs deposited on ITO. Top-view scanning electron microscopy (SEM) and atomic force microscopy (AFM) images of bare ITO and our ETL samples coated on ITO substrates are presented in Figure . It is clear that the morphologies of our ETL samples were quite different from that of bare ITO. Noticeably, ETL samples presented a very uniform, fully covered, and compact surface of SnO2 and SnO2/NDSB-256-4T layers on ITO substrates (Figure b,c). Note that better ETL morphologies will help to form good contact between the ETLs and active layers, as well as help accelerate the electron extraction and transportation processes from the active to the ETL, thus enhancing the overall device performance for iOSCs.[58] The cross-sectional SEM images of SnO2 and SnO2/NDSB-256-4T films coated on ITO substrates are given in Figure S2. Energy-dispersive X-ray (EDX) spectra and elemental mapping analysis for our ETL samples demonstrate good distributions and coexistence of SnO2 and SnO2/NDSB-256-4T ETLs on ITO substrates, as shown in Figures S3 and S4, respectively. As can be clearly seen from Figure S4, the elements C, N, O, S, and Sn of the SnO2/NDSB-256-4T (C12H19NO3S) samples are uniformly distributed on ITO substrates. Note that H is not detectable using normal EDX because of the very low energy of its characteristic radiation. SEM images (Figure b,c) showed that there is not much difference in morphology between the SnO2 and SnO2/NDSB-256-4T. However, a careful observer may see that the SnO2/NDSB-256-4T sample seems to have a very thin and smooth NDSB-256-4T layer on SnO2 compared to the SnO2-only sample (Figure b,c).

Figure 3

Top-view SEM and AFM images of (a,d) bare ITO, (b,e) SnO2, and (c,f) SnO2/NDSB-256-4T ETL samples.

The AFM height images (2 μm × 2 μm) of bare ITO, SnO2 and SnO2/NDSB-256-4T samples are presented in Figure d–f, while the corresponding 3D AFM images are given in Figure S5a–c, respectively. The root mean square values of the roughness (Rq) for the bare ITO, SnO2, and SnO2/NDSB-256-4T samples as displayed in Figure d–f are 1.56, 0.837, and 0.761 nm, respectively. Obviously, both the SnO2 and SnO2/NDSB-256-4T samples revealed lower surface Rq values than bare ITO, demonstrating the excellent morphologies of our ETL samples. Interestingly, the SnO2/NDSB-256-4T sample exhibited slightly lower surface roughness compared to SnO2, suggesting a smoother surface for SnO2/NDSB-256-4T than for SnO2. Note that the smoother surface of ETLs suggests reduced defects, and thus lower recombination rates, a higher current density, and a high fill factor (FF) for iOSCs. We have also further conducted the AFM measurements for ITO, SnO2, and SnO2/NDSB-256-4T samples at a larger area (10 μm × 10 μm); and the results are presented in Figure S6. The Rq values of the bare ITO, SnO2, and SnO2/NDSB-256-4T samples at larger area (10 μm × 10 μm) are found to be 1.62, 1.56, and 1.31 nm, respectively. Obviously, the Rq values of ITO and ETL samples at a larger area (10 μm × 10 μm) are increased as compared to that at a small area size (2 μm × 2 μm) because of some small islands at a larger area on the ETL samples (Figure S6). However, our Rq values at a larger area are still low as compared with other reports,[40] thus demonstrating the excellent morphologies of our ETL samples.

The energy band alignment between each layer of material is considered to be an important factor in the device performance of iOSCs.[59] To understand the important role of energy band alignment of our ETL samples, we used ultraviolet photoelectron spectroscopy (UPS) to determine the valence band maximum (EVBM) energies for SnO2 and SnO2/NDSB-256-4T. Figure a–c presents the UPS survey profiles, valence-band regions (Eonset), and secondary electron cutoffs (Ecutoff) of the SnO2 and SnO2/NDSB-256-4T samples, respectively. Based on the valence-band regions (Eonset) and secondary electron cutoff (Ecutoff) values, the valence band maximum (EVBM) values were −8.22 and −7.55 eV for SnO2 and SnO2/NDSB-256-4T ETLs, respectively. The conduction band minimum levels (ECBM) of the SnO2 and SnO2/NDSB-256-4T ETLs were computed using the energy band gap (Eg) (Figure b) and the valence band maximum (EVBM) values. Conduction band minimum energy levels (ECBM) of the SnO2 and SnO2/NDSB-256-4T ETLs were found to be −4.42 and −3.71 eV, respectively. The WF of the cathode can be calculated using the secondary electron cutoff (Ecutoff) values and the incident photo energy hν = 21.22 eV from a He(I) UPS measurement source.[60] In general, the WF of the cathode can be used in understanding the open open-circuit voltage (Voc) and the electron transport properties of ETLs in iOSCs.[61] The WFs of the SnO2 and SnO2/NDSB-256-4T ETLs were calculated to be 4.42 and 3.85 eV, respectively. It is clear that the WF of SnO2/NDSB-256-4T shifted significantly compared to the SnO2. Thus, iOSCs using SnO2/NDSB-256-4T as ETLs could allow more efficient electron collection to the cathode, and a higher Voc is expected.

Figure 4

(a) UPS surveys, (b) valence-band regions (Eonset), and (c) secondary electron cutoffs (Ecutoff) of SnO2 and SnO2/NDSB-256-4T films deposited on ITO glass substrates. (d) Device structure, (e) energy-level diagram of each layer, and (f) cross-sectional SEM images of the iOSC device using SnO2/NDSB-256-4T ETLs.

Figure d illustrates the device structure of our iOSC devices using SnO2 or SnO2/NDSB-256-4T as ETLs, while the corresponding energy level diagram of each component is presented in Figure e. Compared to SnO2, the conduction band minimum (ECBM) of the SnO2/NDSB-256-4T sample was significantly shifted (−3.71 eV). This value is very close to the lowest unoccupied molecular orbital of PC60BM (−3.70 eV). Because there is no energy offset between the SnO2/NDSB-256-4T ETL and the active layer, photogenerated electrons can be extracted and transferred easily from the active layer to the ITO cathode, resulting in a higher efficiency for iOSCs. The cross-sectional SEM image of a complete iOSC device using SnO2/NDSB-256-4T ETL is given in Figure f. The thickness of the SnO2/NDSB-256-4T ETL was about 30 nm, as can be clearly seen in Figure f. Note that the cross-sectional SEM image of a complete iOSC device using SnO2 as the ETL is shown in Figure S7.

We fabricated P3HT:PC60BM-based standard solar cell devices using SnO2 and SnO2/NDSB-256-4T as ETLs to investigate the effect of NDSB-256-4T interface modification of the SnO2 ETL on the device performance of iOSCs. We optimized the photovoltaic device performance of iOSCs with different NDSB-256-4T concentrations (0, 0.2, 0.5, and 0.8 mg/mL) based on the current density–voltage (J–V) curves, as displayed in Figure a. The dark current and external quantum efficiency (EQE) are presented in Figure S8a,b, respectively. The detailed photovoltaic parameters of each of these iOSCs are summarized in Tables S1–S4. As summarized in Table , the P3HT:PC60BM-based iOSCs using SnO2/NDSB-256-4T (0 mg/mL) exhibited an average PCE recorded over 10 devices of 2.72 ± 0.11%, with a short-circuit current density (Jsc), an open-circuit voltage (Voc), and a FF of 10.20 ± 0.10 mA cm–2, 0.55 ± 0.008 V, and 48.24 ± 1.18%, respectively. Interestingly, the SnO2/NDSB-256-4T samples exhibited excellent performance for iOSC devices up to 0.2 mg/mL concentrations. The device performance of iOSCs started to decrease slightly at a concentration of 0.5 mg/mL, after which the performance was greatly decreased for higher concentrations of NDSB-256-4T (e.g., 0.8 mg/mL). We achieved the average highest photovoltaic device performance (PCE) with high Jsc, Voc, and FF values when an NDSB-256-4T concentration of 0.2 mg/mL was used. The average photovoltaic performance based on 10 devices included a PCE of 3.72 ± 0.04% with Jsc, Voc, and FF values of 10.64 ± 0.14 mA cm–2, 0.60 ± 0.003 V, and 57.90 ± 1.31%, respectively, for the NDSB-256-4T concentration of 0.2 mg/mL. However, a further increase in NDSB-256-4T concentration to 0.5 and 0.8 mg/mL led to reductions in PCE values of 3.56 ± 0.05 and 3.10 ± 0.22%, respectively.

Figure 5

(a) J–V characteristics of the P3HT:PC60BM-based iOSCs with different amounts of NDSB-256-4T (0, 0.2, 0.5, and 0.8 mg/mL)-modified SnO2 ETLs. Device performance of the P3HT:PC60BM-based iOSCs using the bare ITO, NDSB-256-4T (0.2 mg/mL) only, SnO2, and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs: (b) J–V, (c) the dark current, and (d) EQE.

Table 1

Photovoltaic Performance of the P3HT:PC60BM-Based iOSCs Using SnO2/NDSB-256-4T (0, 0.2, 0.5, and 0.8 mg/mL), Bare ITO, NDSB-256-4T (0.2 mg/mL)-Only, and SnO2 as ETLs

ETLs	Jsc (mA/cm2)	Voc (V)	FF (%)	PCE (%)
SnO2/NDSB-256-4T (0 mg/mL)	10.20 ± 0.10	0.55 ± 0.008	48.24 ± 1.18	2.72 ± 0.11
SnO2/NDSB-256-4T (0.2 mg/mL)	10.64 ± 0.14	0.60 ± 0.003	57.90 ± 1.31	3.72 ± 0.04
SnO2/NDSB-256-4T (0.5 mg/mL)	10.33 ± 0.10	0.60 ± 0.001	57.54 ± 0.67	3.56 ± 0.05
SnO2/NDSB-256-4T (0.8 mg/mL)	9.33 ± 0.46	0.60 ± 0.009	55.30 ± 3.78	3.10 ± 0.22
bare ITO	8.84 ± 0.11	0.25 ± 0.010	33.98 ± 0.78	0.74 ± 0.05
NDSB-256-4T (0.2 mg/mL) only	9.71 ± 0.20	0.50 ± 0.013	44.41 ± 1.20	2.16 ± 0.06
SnO2	10.07 ± 0.27	0.58 ± 0.010	47.58 ± 0.98	2.79 ± 0.07

To demonstrate the good device performance of iOSCs based on SnO2 and SnO2 with different concentrations of NDSB-256-4T as ETLs, we also fabricated iOSC devices using bare ITO, NDSB-256-4T (0.2 mg/mL), and SnO2 only as ETLs. The detailed photovoltaic parameters of each of these iOSC devices are given in Tables S5–S7. Figure b–d presents the current density–voltage (J–V) curves, the dark current, and EQE of iOSCs using bare ITO, NDSB-256-4T (0.2 mg/mL), SnO2 only, and SnO2/NDSB-256-4T (0.2 mg/mL) as ETLs for a clear comparison. As summarized in Table , it appears that the average PCEs of the devices based on bare ITO, NDSB-256-4T (0.2 mg/mL), SnO2 only, and SnO2/NDSB-256-4T (0.2 mg/mL) as ETLs were 0.74, 2.16, 2.79, and 3.72%, respectively. These results indicate the prominent role of the electron transport properties of SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) compared to the bare ITO or the NDSB-256-4T (0.2 mg/mL) only.

Electron-only devices are one of the most effective ways to measure the electron-transport properties of the ETLs in OSC devices.[62] Understanding this important aspect, we have also fabricated electron-only devices with a structure of ITO/ETLs/P3HT:PC60BM/LiF/Al to evaluate the electron mobility (μe) when SnO2 or SnO2/NDSB-256-4T (0.2 mg/mL) was used as ETLs. These electron-only devices were measured in the dark with a Keithley 2400 source to obtain the current density–voltage (J–V) characteristics. Figure a,b provides the current density–voltage (J–V) characteristics and the linear fits for the J0.5–V of the electron-only devices using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs, respectively. Following the Mott–Gurney law, which is better known as space charge-limited current (SCLC) theory, we can compute the electron mobility (μe) of these electron-only devices based on the Mott–Gurney expression as[62,63]Here, JSCLC, V, and L are the current density (J), the applied voltage (V), and the thickness of the active layer (P3HT:PC60BM) of the electron-only devices, respectively; ε0 is the permittivity of free space constant, which has a value of 8.854 × 10–12 C V–1 m–1; and εr is the relative dielectric constant of the P3HT:PC60BM (active layer), which usually has a value between 3 and 4 and is typically assumed to be 3.5 as in previous reports.[63,64] After the calculation, the electron mobility values (μe) of the devices using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs were found to be 3.61 × 10–4, and 1.53 × 10–3 cm2 V–1 s–1, respectively. Strikingly, the electron mobility (μe) of the SnO2/NDSB-256-4T (0.2 mg/mL) ETLs was more than four times higher than the devices based on SnO2 only, highlighting the excellent electron-transport properties of devices using SnO2/NDSB-256-4T (0.2 mg/mL) as an ETL. Our electron-only device studies provide convincing evidence of the enhanced photovoltaic performance of iOSC devices with SnO2/NDSB-256-4T (0.2 mg/mL) as ETLs.

Figure 6

(a) Current density–voltage (J–V) characteristics and (b) linear fitting for J0.5–V of the electron-only devices using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs, measured in the dark. (c) Room-temperature PL studies of the active layers coated on SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs. (d) Nyquist plots for the P3HT:PC60BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs, measured in the dark with Vbias = 0 V.

To confirm the impact of NDSB-256-4T modification on SnO2 ETLs with regard to the charge extraction and recombination, we also studied the PL of the glass/SnO2/P3HT:PC60BM and glass/SnO2/NDSB-256-4T (0.2 mg/mL)/P3HT:PC60BM samples. As displayed in Figure c, the glass/SnO2/NDSB-256-4T (0.2 mg/mL)/P3HT:PC60BM sample exhibited significant PL quenching compared to the glass/SnO2/P3HT:PC60BM. This suggests greatly reduced recombination rates of electrons and holes and enhanced electron extraction and transportation ability. The reason for this PL quenching likely originated from successful passivation of surface trap states in glass/SnO2/P3HT:PC60BM with the presence of NDSB-256-4T modification, significantly reducing the oxygen vacancy-related defects in SnO2. Therefore, the PL results again confirmed the enhancements in device performance of iOSCs using NDSB-256-4T (0.2 mg/mL)-modified SnO2 ETL.

Electrochemical impedance spectroscopy (EIS) was also studied to investigate the charge transport behaviors at the interface between the ETL and the active layer of iOSC devices.[65−67] In general, the interpretation of the impedance spectra (IS) is directly related to the applied bias voltage for the iOSC devices.[66] When one conduct IS measurements at the open-circuit, the photovoltaic device will mainly work under recombination conditions. In this case, the IS is usually characterized by a major RC arc along with additional minor features at high frequency.[66] The high-frequency part of the spectra (minor RC arc) may provide information of intrinsic series resistances in iOSC devices, while the low frequency arc (major RC arc) is ascribed to recombination in the active layer.[66] This typical IS can be found in previous reports.[43,68] Because the recombination resistance increases exponentially as the applied voltage decreases, therefore, at low applied voltages, the IS response becomes represented for the intrinsic series resistances in iOSC devices in the equivalent circuit like the sheet resistance (RS) and the charge transfer resistance (RCT), which are effectively voltage independent rather than for the recombination resistance.[66,69] Note that the charge transfer resistance at the low voltage can also be used as the useful source for understanding the recombination resistance in the photoactive blend: low charge transfer resistance might suggest that higher recombination resistance in the photoactive blend and vice versa. For IS measurements conducted in the dark with an applied voltage V = 0, the response is usually characterized by only one RC arc that contains information for transport and series resistance elements as in several previous reports.[65,67,70] In our iOSCs, we have used the same photoactive blend along with the same geometry device except for the ETLs (SnO2 vs SnO2/NDSB-256-4T), thus the RCT resistances (resistance at the active/ETL interface, and the resistance at the ETL/ITO interface) are crucial important to evaluate the electron extraction and transport properties of SnO2 and SnO2/NDSB-256-4T. Because we measured IS in the dark with an applied voltage V = 0, thus our IS should be represented for intrinsic series resistances of our iOSCs rather than for the recombination resistance in the photoactive blend, as well-consisting with several previous reports.[65,67,70]Figure d displays the Nyquist plots for the P3HT:PC60BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs measured in the dark with Vbias = 0 V. The equivalent circuit for these Nyquist plots is given in the inset of Figure d. The intrinsic series resistances of our iOSCs can be divided into two components. One is the RS and the other is the RCT, which consists of the internal resistance of the active layer, the resistance at the active/ETL interface, and the resistance at the ETL/ITO interface.[65,67,70]Figure d shows that the RCT values for iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs were 1718.60 and 438.04 Ω cm2, respectively. Obviously, the RCT of the SnO2/NDSB-256-4T (0.2 mg/mL)-based iOSC devices showed a dramatic decrease down to 438.04 Ω cm2 compared to the SnO2-only devices (1718.60 Ω cm2). This indicates that NDSB-256-4T-modified SnO2 may result in a reduction in the active/ETL interface resistances, which consequently accelerates electron transport/transfer between the active layer and the ETL. The EIS results also provide compelling evidence that can be used to explain our iOSC device performance.

Among the three essential characteristics for any photovoltaic technology (efficiency, scalability, and stability), obtaining good stability is one of the most difficult challenges hindering the eventual commercialization of iOSCs. Considering this, we also studied the stability of our iOSC devices. The iOSC devices (without any encapsulation) were kept in ambient conditions, and the photovoltaic performance was regularly checked over 5 weeks. Figure demonstrates the results of the stability studies for our iOSC devices, noting that the detailed photovoltaic parameters of these devices were also recorded and are presented in Tables S8 and S9 for SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL), respectively. Remarkably, the iOSCs using SnO2/NDSB-256-4T (0.2 mg/mL) showed outstanding long-term stability with a PCE that remained at over 98% of its initial values after 5 weeks. The inverted device structure, the ultra-stability of SnO2, and the outstanding interface engineering improvement between the active layer and the SnO2 with the presence of the NDSB-256-4T modification equally contributed to the excellent long-term device stability of our iOSCs.

Figure 7

Device stability studies for the P3HT:PC60BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs: (a) PCE, (b) Jsc, (c) Voc, and (d) FF.

In a further application of the PTB7-Th:PC70BM systems, we also obtained impressive enhancements in device performance for iOSCs using SnO2/NDSB-256-4T (0.2 mg/mL) as the ETL. Figure a–c presents the photovoltaic performance of PTB7-Th:PC70BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2, 0.5, and 0.8 mg/mL) ETLs, while the corresponding EIS studies are also given in Figure d. The photovoltaic parameters are summarized in Table , while the detailed photovoltaic parameters for each of these iOSC devices are provided in Tables S10–13. As shown in Table , the average PCEs of the devices with SnO2 and SnO2/NDSB-256-4T (0.2, 0.5 and 0.8 mg/mL) ETLs were 4.45, 8.22, 8.06, and 7.69%, respectively. Compared to the devices based on SnO2 only, the average PCEs of the devices using NDSB-256-4T (0.2, 0.5, and 0.8 mg/mL)-modified SnO2 as ETLs showed dramatic improvements. This improvement in photovoltaic performance is attributed to the considerably smaller leakage current (Figure b) along with the significantly reduced charge transfer resistance (RCT) for iOSC devices using NDSB-256-4T (0.2, 0.5, and 0.8 mg/mL)-modified SnO2 as ETLs.

Figure 8

Device performance of the PTB7-Th:PC70BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2, 0.5, and 0.8 mg/mL) ETLs: (a) J–V, (b) the dark current, (c) EQE, and (d) Nyquist plots.

Table 2

Photovoltaic Performance of the PTB7-Th:PC70BM-Based iOSCs Using SnO2 and SnO2/NDSB-256-4T (0.2, 0.5, and 0.8 mg/mL) as ETLs

ETLs	Jsc (mA/cm2)	Voc (V)	FF (%)	PCE (%)
SnO2	15.03 ± 0.16	0.64 ± 0.012	46.23 ± 0.93	4.45 ± 0.12
SnO2/NDSB-256-4T (0.2 mg/mL)	15.49 ± 0.29	0.78 ± 0.002	68.24 ± 0.89	8.22 ± 0.10
SnO2/NDSB-256-4T (0.5 mg/mL)	15.27 ± 0.22	0.78 ± 0.004	67.93 ± 0.72	8.06 ± 0.10
SnO2/NDSB-256-4T (0.8 mg/mL)	15.20 ± 0.08	0.78 ± 0.005	65.07 ± 1.34	7.69 ± 0.15

We have also studied the stability of the PTB7-Th:PC70BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) as ETLs. The iOSC devices (without any encapsulation) were stored in ambient conditions, and the photovoltaic performance was regularly checked over 4 weeks. The device stability results of the PTB7-Th:PC70BM-based iOSCs using SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) as ETLs are presented in Figure S11; while the detailed photovoltaic parameters of these devices were also recorded and are gathered in Tables S14 and S15 for SnO2 and SnO2/NDSB-256-4T (0.2 mg/mL) ETLs, respectively. As expected, the PTB7-Th:PC70BM-based iOSCs using SnO2/NDSB-256-4T (0.2 mg/mL) ETLs revealed excellent long-term stability with PCE remained at over 90% after being stored for 4 weeks in ambient conditions. Again, the high-stability of SnO2 in the inverted device structure and the excellent interface engineering improvements between the active layer and the SnO2 with the presence of the NDSB-256-4T modification are compelling reasons that explain the excellent long-term device stability of our iOSCs.

Conclusions

In summary, we successfully introduced a promising interfacial engineering strategy to achieve highly efficient and stable iOSCs using a low-temperature solution-processed zwitterion NDSB-256-4T-modified SnO2 ETL. We demonstrated that NDSB-256-4T helps both to reduce the WF and passivate the defects in SnO2. Therefore, better interfacial contact between SnO2 and the active layer was built, resulting in a higher PCE and longer device stability of iOSCs using SnO2/NDSB-256-4T ETL compared to devices based on SnO2 only. We achieved an average PCE of 3.72% for the P3HT:PC60BM-based iOSCs using SnO2/NDSB-256-4T (0.2 mg/mL) as the ETL; this is 33% higher than devices using SnO2 only (2.79%). Our iOSC devices also showed excellent stability, with 90% of the PCE remaining after storing 5 weeks in ambient air without encapsulation. In an extended application for the PTB7-Th:PC70BM systems, we also obtained an impressive average PCE of 8.22% with SnO2/NDSB-256-4T (0.2 mg/mL) as the ETL. Devices based on SnO2 exhibited an average PCE of only 4.45%. We believe that the use of NDSB-256-4T to modify SnO2 ETL is an effective way to obtain highly efficient and stable commercial iOSCs. Finally, we suggest that the SnO2/zwitterion ETL approach can be applied to other optoelectronic devices such as organic light-emitting diodes or PSCs, where an ETL is necessary to ensure good device performance.

Experimental Section

Materials and Reagents

Tin(II) chloride dihydrate (SnCl2·2H2O, ∼98%) was received from Alfa Aesar, and ethanol (C2H5OH, 99.9%) was purchased from EMD Millipore. Tin(II) chloride dihydrate and ethanol were used as precursor and solvent, respectively, to prepare the SnO2 precursor solution. Methanol (CH3OH, 99.8%) and 3-(4-tert-butyl-1-pyridinio)-1-propanesulfonate (NDSB-256-4T, 98%) were purchased from Sigma-Aldrich to prepare the NDSB-256-4T solution.

The donor polymers included poly(3-hexylthiophene-2,5-diyl) (P3HT) and poly[4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)benzo[1,2-b;4,5-b′]dithiophene-2,6-diyl-alt-(4-(2-ethylhexyl)-3-fluorothieno[3,4-b]thiophene-)-2-carboxylate-2-6-diyl] (PTB7-Th), which were purchased from Rieke Metals, Inc. and Solarmer Energy Inc., respectively. Both acceptor polymers, PC60BM (phenyl-C60-butyric acid methyl ester) and PC70BM (phenyl-C70-butyric acid methyl ester), were obtained from Nano Holding. The ODCB (1,2-dichlorobenzene, 99%) and DIO (1,8-diiodooctane, 98%, contains copper as stabilizer) were purchased from Sigma-Aldrich to prepare the donor–acceptor blend solution.

Fabrication of the iOSC Devices

ITO-glass substrates (2.5 × 2.5 cm2) with a sheet resistance of about 12 Ω/square and a thickness of 125 nm were obtained from the Korea Electronics Technology Institute and were used as cathode electrodes of iOSC devices. Prior to coating the ETLs, the ITO-glass substrates were well-cleaned through sequential ultrasonic treatments with deionized water, acetone, and isopropanol. Low-temperature solution-processed SnO2 ETLs were prepared according to our previous reports.[71] SnCl2·2H2O (1.128 g) was dissolved in 50 mL of absolute ethanol. Then, the resulting solution was stirred at 80 °C for 8 h to form a uniform solution. After aging for 1 day, this solution was used to prepare SnO2 ETL (with a thickness of 20–25 nm) via a spin-coating method (3000 rpm, 40 s) onto either a bare glass or ITO-glass substrates in air. Then, they were sintered on a hotplate at 185 °C for 1 h. Different concentrations of NDSB-256-4T (0, 0.2, 0.5 and 0.8 mg/mL) were prepared by dissolving 0, 2, 5, and 8 mg of NDSB-256-4T in four different vials, in which each contained 10 mL methanol. Before spin-coating, these vials were stirred at room temperature for about 3 h to yield uniform solutions. The deposition of NDSB-256-4T films (∼8–10 nm) took place inside a N2-filled glovebox via a spin-coating method at 4000 rpm for 40 s, followed by drying at 100 °C for 10 min using a hotplate.

The blend active layers (P3HT:PC60BM) in ODCB (1,2-dichlorobenzene) with a concentration of 25 mg/mL were spin-coated on ETLs (500 rpm, 40 s) to yield an active thickness of about 180–200 nm. Then, these freshly coated active samples were kept in closed petri dishes overnight to allow a gradually solvent annealing process, as in previous reports.[72] Meanwhile, PTB7-Th:PC70BM active layers with the same thickness of 180–200 nm were obtained by spin-coating a blended PTB7-Th:PC70BM solution (1:1.5 wt %, 30 mg/mL) in mixed solvent ODCB/DIO (97:3 vol %) on ETL samples at a speed of 500 rpm for 40 s. PTB7-Th:PC70BM samples were kept in a vacuum chamber for ∼3 h prior to electrode deposition processes to completely dry the films and remove the additive (DIO).[73] Finally, the iOSC devices were thermally deposited with a 10 nm MoO3 HTL, followed by a 100 nm Ag anode electrode through a shadow mask in a high vacuum chamber (less than 1 × 10–6 Torr). The active area of the iOSC device was 0.11 cm2 as determined using an aperture shadow mask. The thermal evaporation rates were controlled at 0.1–0.2 and 2–2.5 Å/s for MoO3 and Ag, respectively. For the electron-only devices, the anode contacts of LiF (0.7 nm) and Al (105 nm) were also thermally deposited on the active layer (P3HT:PC60BM) via a shadow mask with controlled thermal evaporation rates of 0.1 and 2–3 Å/s for LiF and Al, respectively.

Thin-Film Characterization

Shimadzu UV-2550 spectrophotometer systems were first used to evaluate the optical properties (UV absorption and transmittance) of SnO2 and SnO2/NDSB-256-4T ETLs on bare glass substrates. The room-temperature PL spectra of the ETL and the active samples were achieved under excitation at 350 nm via a Jasco FP-6500 spectrophotometer system. The XRD and XPS systems were used to obtain the structural properties and chemical states of SnO2 and SnO2/NDSB-256-4T samples, respectively. The band energy levels of SnO2 and SnO2/NDSB-256-4T samples were investigated using UPS measurements, with an energy source of 21.22 eV (He I). Field-emission SEM (FESEM, Zeiss Co., Germany) and AFM with tapping mode (Bruker, USA) were used to obtain the morphology information of ETLs. The EDX (energy-dispersive X-ray spectroscopy) of the SEM system was used to obtain elemental maps of SnO2 and SnO2/NDSB-256-4T samples. The cross-sectional SEM images of SnO2 and SnO2/NDSB-256-4T samples along with the complete iOSC devices using SnO2 or SnO2/NDSB-256-4T, as ETLs were also achieved by using the FESEM measurement system (Zeiss Co., Germany).

Photovoltaic Characterization

The photovoltaic device performances of all fabricated iOSCs were tested under standard illumination (100 mW/cm2 AM 1.5 G) or in the dark under ambient conditions with the solar cell I–V simulator measurement system using Keithley 2400. For normal photovoltaic characterization, the intensity of the light illumination was calibrated before the measurements using a monocrystalline-silicon solar cell (2 × 2 cm2, calibrated at NREL, Colorado, USA) to set the standard conditions (100 mW/cm2 AM 1.5 G). We scanned all the iOSC devices in the reverse direction with a scan speed of 0.2 V/s. The delay time was 10 ms; the scan step was 0.02 V from −1 to 1 V. The electron-only devices were also characterized using Keithley 2400 with an applied voltage ranging from 0 to 3 V, measured in the dark. A scan step of 0.05 V at a scan speed of 0.2 V/s and a delay time of 100 ms were used. EQE was obtained using a Polaronix K3100 spectrometer system. Finally, the impedance measurements for all the iOSC devices were performed using a VersaSTAT3 (METEK) tool conducted in the dark with an applied voltage Vbias = 0 V, and the frequency was selected between 10 kHz and 0.1 Hz.