Because of the tunable bandgap, solution processability, and modifiable surface [1], quantum dots (QDs) are now widely used in light-emitting diodes [2, 3], solar cells [4], field-effect transistors [5], photodetectors [6], and so on. In the field of solar cells, near-infrared QDs have become an ideal bottom cell absorber for tandem devices due to their quantum size effect, low-cost solution fabrication, and high infrared absorption coefficient [7,8,9,10,11,12]. The theoretical research shows that the tandem device consisting of a 1.55 eV perovskite top-cell and 1.0 eV PbS QDs bottom-cell can achieve an ideal power conversion efficiency (PCE) of 43%, which is much higher than the 33% single junction limit [13]. In comparison with all perovskite tandem solar cells, such a tandem device also holds special advantages of high working stability and low production cost. Thus, QD near-infrared solar cells (IRSCs) show high potential in next-generation photovoltaics and tandem solar cells.

For thin film solar cell fabrication, it is very important to produce homogeneous, dense, and large-area films. Nowadays, benefiting from the development of the solution-phase ligand exchange method [14], high-quality QD films can be prepared by a single step of spin-coating. Employing a high-speed gas flow can result in rapid drying of the solvent, to obtain a smooth and continuous QD film. We have previously reported the sputtered ZnO-based PbS QD devices by spin-coating (Eg = 0.97 eV) and realized 3.90% and 10.47% certified PCE under 800 nm filter and AM1.5G conditions, respectively [15]. However, it is worth noting that the spin-coating process always wastes a significant quantity of QD ink, which increases the manufacturing cost greatly. Moreover, spin-coating is a small-area deposition method and is not conducive to the large-scale manufacturing of QD solar cells and this is a major obstacle to commercial development.

Blade coating without centrifugal force is expected to solve these problems and achieve low-cost, large-area preparation of QD films [16]. Sukharevska et al. have demonstrated the preparation of stable QD ink and large-scale film by blade coating with a PCE of 8.7% [17]. The development of QD ink film production from the mature spin coating process to the blade coating process still faces many physical and chemical challenges. The selection of solvents is an important factor in preparing high-quality films by blade coating [16,17,18]. Specifically, the solvent physical parameters should match the blade coating process. The suitable solvent can prolong the effective window period of film preparation, and ink with a specific viscosity can prevent the autoflow of the wet film to reduce macro defects. At the same time, the solvent needs to have a suitable boiling point to avoid uneven film caused by fast solvent evaporation [17].

Here, we developed a mixed solvent system from dimethylformamide (DMF) and butylamine (BTA) according to Lewis acid–base theory and DLVO theory. This system can maintain the stability of QD inks and is compatible with the blade coating process. The stability time of QD ink based on this system exceeded 7 h, and 100 cm2 uniform QD films were successfully obtained by blade coating. The large area QD film based devices achieved an average PCE of 11.14% and an 800 nm-filtered PCE of 4.28%, which are the top values achieved in reported research literature to date for the blade coating method.

2.1 Chemicals

Thioacetamide (TAA) (≥ 99%), zinc stearate (≥ 90%), 1-octadecene (ODE) (Aladdin, ≥ 90%), oleylamine (OLA) (Aladdin, 90%), ethanol (Sinopharm, ≥ 99.7%), lead chloride (PbCl2) (Aladdin, 99.99%), acetone (Sinopharm, ≥ 99.5%), oleic acid (OA) (Alfa Aesar, ≥ 90%), n-octane (Sinopharm, ≥ 95%), hexylamine (Aladdin, 98%), hexamethyldisilathiane (TMS) (TCI, ≥ 95%), DMF (Aladdin, ≥ 99.8%), BTA (Aladdin, ≥ 98%), indium tin oxide (ITO) (Advanced Election Technology Co. Ltd.), lead bromide (PbBr2) (Advanced Election Technology Co. Ltd., ≥ 99.99%), lead iodide (PbI2) (Advanced Election Technology Co. Ltd., ≥ 99.999%), ethyl acetate (Sinopharm, ≥ 99.7%), Dimethyl sulfoxide (DMSO) (Aladdin, ≥ 98%), N-Methyl-2-pyrrolidone (NMP) (Aladdin, ≥ 99%), Pyridine (Py) (Aladdin, ≥ 98%), Propylene carbonate (PC) (Aladdin, ≥ 98%), 1, 2-ethanedithiol (EDT) (Aladdin, ≥ 97%) and acetonitrile (Sinopharm, ≥ 99.8%).

2.2 Synthesis of small-size PbS QDs (exciton peak ~ 880 nm)

Lead oxide, oleic acid, and ODE were placed in a 250 mL three-necked flask and heated under a negative pressure condition for 1.5 h, to obtain lead oleate solution. Then, under intense agitation, the TMS was injected into the lead oleate solution. One minute later, the three-necked flask was bathed to room temperature to generate small-size quantum dots. PbS QDs purified with ethyl acetate and ethanol were finally dispersed in n-octane (40 mg/mL) for EDT layer deposition.

2.3 Synthesis of ZnS QDs (absorption peak ≈ 256 nm)

3.6 g thioacetamide (0.048 mol), 70.56 g zinc stearate (0.096 mol), 221 g ODE and 156 g OLA were put into a 2 L three-neck flask. The mixture was heated to 140 °C under a nitrogen atmosphere and maintained for 50 min. Then, the water bath, was cooled to 40 ℃ and 32 mL n-octylamine was added. Finally, n-hexane and ethanol were used as antisolvents to precipitate ZnS QDs [27].

2.4 Synthesis of large-size PbS QDs (exciton peak ≈ 1290 nm)

1.946 g PbCl2 (0.007 mol) and 18.94 g OLA were placed into a 250 mL three-mouth flask. The mixture was heated to 140 ℃ under a nitrogen atmosphere and maintained for 30 min. Then the flask was cooled to 60 °C and ZnS QDs were rapidly injected for nucleation. After that, ZnS QDs were slowly injected for growth (absorbance was 0.3 after 3000 times dilution). The growth temperature gradually increased from 60 °C to 100 °C, and the whole growth process lasted for about 90 min. Then, the temperature was cooled to 70 °C by water bath, 90 mL hexane was injected, and 40 mL oleic acid was injected at 40 °C. Finally, the mixture was purified with acetone to obtain solid PbS QDs [28].

2.5 Preparation of solar cells

Firstly, the ZnO layer was deposited by sputtering on a glass-ITO substrate. Secondly, the absorption layer was prepared by scraping. Specifically, the ligand solution of PbI2:PbBr2 was prepared with 10 mL DMF, and the molar ratio of PbI2 to PbBr2 was 9.2:1. An equal volume of 10 mg/mL QDs solution in n-octane was then prepared. The above two were thoroughly mixed for ligand exchange, and the exchanged solid QDs were obtained by centrifugation. Then 700 mg/mL QD ink was prepared with two mixed solvents (BTA:DMF = 4:1 and BTA:DMF = 3:7) for blade coating. The QD films were deposited by blade coating on the ZnO substrate at 40 °C. The width of the slit was 90 μm and the coating speed was 10 mm/s. The wet film was annealed at 90 ℃ for 10 min to obtain a dry QD film with a thickness of about 350 nm. The whole process was carried out under a nitrogen atmosphere. Thirdly, PbS QD film (exciton peak ~ 880 nm) was treated with EDT ligand used as hole transport layers. Finally, 80 nm gold was deposited on the top as the top electrode. The effective area of the solar cell was 0.04 cm2.

2.6 Materials and device characterizations

The optical absorption spectra of the QDs were measured by a Shimadzu UV-3600 Plus spectrophotometer. Photoluminescence (PL) spectra of QDs were measured using HORIBA modular scientific research grade fluorescence spectrometer. The QD film surface topography was measured by atomic force microscope (AFM) SPM9700. The scanning electron microscope (SEM) images were obtained by FEI Nova Nano SEM 450. The PbS QD film crystallization was tested by X-Ray diffraction (XRD) with Cu Kα radiation (Philips, X pert pro-MRD, Netherlands). The current density–voltage (J–V) curve was obtained under a simulated AM 1.5 (100 mW/cm2) solar spectrum from a 450 W xenon lamp (Oriel, Model 9119, Newport). 800 (Thorlabs FELH-0800) and 1100 nm (Thorlabs FELH-1100) long pass filters were used to simulate the four-terminal tandem configurations with perovskite and silicon, respectively. Capacitance–voltage (C–V) measurements were conducted using an Agilent 4200A at a frequency of 100 kHz and an AC signal of 50 mV, scanning from − 0.7 to 0.7 V, in steps of of 20 mV. The drive-level capacitance profiling (DLCP) measurement of the devices was performed with variant amplitude (≈ − 0.7–0.6 V) and frequency (10–500 kHz). The transient photocurrent (TPC) and transient photovoltage (TPV) measurements were performed on the device under dark conditions. A ring of red light-emitting pulse diode (LED, Lumiled) was controlled by a fast solid-state switch, and the pulse width was 1 ms. The TPC was measured using 40 Ω external series resistance to operate the device in short-circuit conditions. Similarly, TPV was applied using 1 MΩ external series resistance to operate the device in open-circuit conditions. Images of QD aggregates were obtained using a transmission electron microscope (TEM) instrument (JEOL 100FJEM-2100F). The external quantum efficiency of the solar cell was obtained using a McScience K3100IQX measurement system (100 Hz chopper monochromatic illumination).

In order to make QD ink compatible with blade coating, the ink solvent recipe was studied. The stability of PbS-IBr (PbI2 and PbBr2 capped PbS QDs) ink depends on the electrostatic interaction between the charged surface of QDs and the surrounding solvent. The PbS (111) facet is lead-rich [19], for dissolving the QDs, solvents need to have sufficient capacity to provide lone pair electrons; that is, sufficient Lewis basicity can help to dissociate ligand precursors. To study the exact effect of solvent on the ink, I/Br capped PbS QDs were dissolved in typical basic solvents; there were BTA, NMP, PC, Py, etc. As shown in Fig. S1a, QDs only dispersed in BTA due to its strong Lewis basicity. Key physical parameters for the above solvents are listed in Table S1. However, BTA does not maintain the stability of the ink. As shown in Fig. S1b, for BTA-based ink, no obvious exciton peak of QDs was obtained after 4 h of storage, indicating the degradation of QDs.

The instability of BTA-based ink was analyzed by DLVO theory. (Details of the theory are given in the supporting information [20], Additional file 1.) Fig. 1a shows the relationship between the potential energy of colloidal particles and spatial spacing. And the potential energy (V) is the sum of gravitational potential energy (Va) and repulsive potential energy (Vr). When the dielectric constant of the solvent is small, the corresponding V is very small, and has difficulty in impeding QD aggregation. As shown in Fig. 1b, c, there is a serious overlap in the diffusion layer of the colloidal particles, resulting in colloidal instability. On the contrary, when the dielectric constant of the solvent is large, as shown in Fig. 1d, there is a much higher potential energy compared with the scenario of Fig. 1a. As shown in Fig. 1e, f, high potential energy can keep a low degree of overlap in the diffusion layer and maintain the stability of the colloid. Thus, the lowest dielectric constant of BTA ~ 4.9 in our investigation was not sufficient to maintain the stability of the ink. Moreover, the boiling point of BTA at standard atmospheric pressure is only 78 ℃, which makes it evaporate rapidly at room temperature and it is incompatible with the large-area preparation process [21, 22]. On the other hand, solvents with boiling points that are too high are difficult to remove in subsequent annealing processes. Therefore, an optimal solvent should have strong Lewis basicity, a high dielectric constant, and a suitable boiling point. It is difficult for a single solvent to meet all of the above requirements, thus we adopted a mixed solvent strategy to address this challenge.