Improved Perovskite/Carbon Interface through Hot-Pressing: A Case Study for CsPbBr3-Based Perovskite Solar Cells.

Due to the low cost and printable nature of the carbon paste, carbon-based perovskite solar cells (PSCs) are attractive for real application. However, the poor contact at the perovskite/carbon interface obviously hinders the achievable fill factor of the carbon-based PSCs. In this work, we introduce a pressure-assisted method to improve the contact at the perovskite/carbon interface. Via modulating the applied pressure, the power conversion efficiency of CsPbBr3 PSCs (small area) can be improved from the initial 7.40% to 7.95% (pressing) and 8.34% (hot-pressing). A more remarkable feature is that the hot-pressing process boosted the performance from 5.1% (normal) to 6.9% (hot-pressing assisted) of large-scale (0.5 cm2) devices, a more than 30% enhancement. Finally, the hot-pressing method introduced in this work shows great prospects for improving the efficiency of carbon-based PSCs, especially large-scale PSCs.

Introduction

Benefited from their excellent optical and electrical properties, organic–inorganic metal halide perovskite solar cells (PSCs) have attracted great attention among both academic and industrial fields.[1−5] Meanwhile, conventional PSCs use noble 2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD) and costly metal electrodes, which reduce the cost-effectiveness of PSCs. Carbon-based PSCs, which exclude hole transport materials and replace the metal electrode with carbon paste, turn out to be an alternative option. To start with, Han et al. first demonstrated the carbon-based PSCs with a device structure of FTO/TiO2/ZrO2/perovskite/carbon.[6] Yang et al. later suggested that candle soot can be an excellent hole extraction layer for PSCs.[7] Subsequently, multiple carbon materials including graphite,[8−10] carbon black,[11,12] spongy carbon,[13] and carbon cloth[14] are developed for PSC application. Very recently, the power conversion efficiencies (PCEs) of carbon-based HTL-free PSC have been boosted to 18.9%,[15] and fully printed planar carbon-based PSCs shows a record PCE of 15.3%.[16]

Although progressive achievements have been reached for carbon-based PSCs, the record efficiency (18.9%) is far lower than that of conventional PSCs (25.5%).[17] This is largely due to the poor contact at the perovskite/carbon interface. To be more specific, because the carbon paste is composed of a graphite flake and carbon sphere, pinholes are unavoidable at the perovskite/carbon interface. This results in pronounced photo-carrier recombination. Several researchers have demonstrated a better adhesion at the perovskite/carbon interface-promoted carrier transportation.[18−25] Then, interface engineering turns out to be an effective strategy for enhancing the performance of carbon-based PSCs. Materials including inorganic compounds (MnS[26] and Mxene[27]) and organic materials (PEDOT, PPy, PANi,[28] CuPc,[29] hexamethylenetetramine,[30] and 1-butyl-2, 3-dimethylimidazolium chloride ([BMMIm]Cl) ionic liquids (ILs)[31]) are introduced to improve the interface contact and promote charge extraction. Yang et al. introduced cesium acetate (CsAc) at the perovskite/carbon interface, contributing to an improved VOC.[25] What is more, formula engineering of the carbon paste also results in a better contact of the perovskite/carbon interface.[32−36] Although the aforementioned methods are attractive, they do not change the fluffy nature of the carbon layer.

In this study, we propose a simple hot-pressing method to solve the poor contact issue at the perovskite/carbon interface. CsPbBr3 has been selected as the photoactive layer because the PCEs of carbon-based CsPbBr3 PSCs are comparable or even superior to those of conventional metal electrode-structured devices.[37−39] After pressure treatment, the pinhole-free perovskite/carbon interface with reduced carbon layer thickness suggests an enhanced interfacial contact, benefiting the carrier extraction. We receive a PCE of 8.34% for hot-pressing-treated PSCs, compared with 7.4% of the control devices. Moreover, the method works better for large scale devices. PCE of large-scale devices has been increased by ∼30% (5.1% to 6.9%) after the hot-pressing treatment. Finally, thanks to the dense carbon layer after the hot-pressing process, the hot-pressing-treated PSCs also show improved stability at ambient conditions. After all, the hot-pressing method seems to be an effective method for developing high-efficiency, reproducible, and stable carbon-based PSCs.

Results and Discussion

We start by describing the device fabrication process. The main difference of the device fabrication lies in the deposition of the carbon electrode. For the control devices, the commercial carbon paste is doctor-blade coated on the CsPbBr3/SnO2/FTO substrates and heated at 100 °C for 20 min. The pressing process refers to applying pressure to the fabricated devices at ambient conditions (Figure G1). Meanwhile, the hot-pressing process suggests adding pressure on devices during the annealing process of the carbon paste (Figure G2). The experimental details are listed in the Supporting Information. The influence of the G1 and G2 processes will be discussed later.

Figure 1

Schematic diagram of (G1) the pressing process and (G2) the hot-pressing process.

As for the CsPbBr3 film fabrication, we use the methanol process developed by Tang et al.[33,40,41] The phase transition between CsPb2Br5, CsPbBr3, and Cs4PbBr6 can be tuned by adjusting the spin-coating cycle of the CsBr precursor. With an optimized spin-coating cycle (8), the prepared films show diffraction peaks at 2θ = 15.2°, 21.6°, 30.7°, and 44.1° on the XRD spectrum, corresponding with the (100), (110), (200), and (220) lattice planes of the CsPbBr3 phase, respectively (Figure S1). No additional peak showed up, which suggested the phase pure nature of the prepared films. The UV–vis absorption spectrum shows a distinct absorbance at ∼540 nm. This also supports the high quality of the prepared CsPbBr3 films (Figure S2).

The SEM images illustrate that the prepared films are polycrystalline with a crystal size of ∼2 μm (Figure a). Grain boundaries are distinct, and pinholes can hardly be visible, which support the high quality of the prepared films. The average thickness of the prepared films is about 400–500 μm, consistent with the previous research (Figure b).[42] We are more interested in the morphological transformation of the carbon electrode layer before and after pressing. Listed in Figure c, the thickness of the carbon electrode layer for the control devices is ∼16 μm, whereas this has been reduced to 11 μm for the hot-pressing devices, an almost 1/3 shrinkage. Moreover, shown on the side-view SEM images, there are obvious cracks and un-touched areas at the perovskite/carbon interface (marked in red), pointing to a poor contact (Figure c). Meanwhile, the hot-pressing carbon films show that the graphite flakes in the carbon paste are horizontally aligned, and the carbon electrode and the perovskite layer are tightly connected (Figure d); thus, effective carrier transportation can be guaranteed.

Figure 2

(a) Top-view SEM images of the CsPbBr3 films deposited on the SnO2/FTO substrate. (b) Cross-sectional SEM images of the CsPbBr3 films. Cross-sectional SEM images of (c) the control device and (d) the hot-pressing device.

Next comes to the performance characterization. Under light illumination, the generated photo-electrons flow from the photo-active layer (CsPbBr3) down to SnO2 and are collected by the FTO substrate. The reverse is the case for the photo-holes (Figure a). As shown in Figure b and Table S1, the control device received an open-circuit voltage (VOC) of 1.45 V, a short circuit current (JSC) of 6.80 mA cm–2, a fill factor (FF) of 0.75, and an overall power conversion efficiency (PCE) of 7.4%. This PCE has been increased from 7.40% to 7.93% (pressing) and 8.34% (hot-pressing). The integrated photocurrents calculated from the incident photon-to-electron conversion efficiency (IPCE) are 6.50 mA cm–2 (control) and 6.80 mA cm–2 (hot-pressing treated). The results are in good agreement with the JSC measured from the J–V characterization (Figure c, Table S1). After introducing the pressure-assisted process, the performance enhancement is mainly due to an improved JSC (from 6.80 mA cm–2 to 7.18 mA cm–2) and FF (from 0.75 to 0.79), corresponding well with a better contact at the selective interface. As for the reproducibility of the devices, 20 devices are selected for control and hot-pressing-treated process separately. The corresponding statistical histogram is provided in Figure d. The control devices yield a PCE of only 5–7%. In comparison, most of the hot-pressing-treated devices receive a PCE over 7%. Meanwhile, photocurrent hysteresis for the control and hot-pressing-treated devices is analyzed by the J–V curves (Figure S3 and Table S2). Following previous reports,[43] photocurrent hysteresis can be expressed by the photocurrent hysteresis index (HI), which can be calculated as follows:

Figure 3

(a) Band alignment diagram of the CsPbBr3 PSCs. (b) J–V characterization of the PSCs. The J–V curves have been measured at 0.1 V/s. (c) IPCE spectrum and the integrated JSC calculated from the IPCE spectrum of the prepared CsPbBr3 PSCs. (d) Histogram of the PCE values for PSCs prepared from different processes. Twenty devices were collected.

The HI of the hot-pressing-treated devices is 6.8%, which is lower than that of the control devices (14.9%), indicating a reduced defect and an enhanced contact at the interface between the carbon and perovskite layer. Besides, the hot-pressing-treated devices also show a narrow PCE distribution (Figure d). These results indicate the advantages of the hot-pressing process.

We are more interested in if the pressing process can be applicable for large scale devices. A 0.5 cm2 device has been selected, and the relating PCEs are recorded in Figure a,b and Tables S3 and S4. The control process only exhibited a PCE of 5.13%, whereas this has been increased to 6.02% in the pressing-assisted process. The hot-pressing process further promotes the PCE to 6.93% with a JSC of 6.72 mA cm–2, VOC of 1.37 V, and FF of 0.76. This is about 30% higher than that of the control process. The results suggest that the hot-pressing process is more effective for large-scale devices. It is more interesting to note that the PCE of the pressing-treated PSCs dropped after 24 h of storage, whereas the PCE for the hot-pressing-treated PSCs remained almost constant during a 24 h storage (Figure ). The influence of the applied pressure on the PCE of large-scale devices is listed in Figure S4 and Table S5, and the value of the optimum pressure is 32 N cm–2. The achievable PCE first rises with the increased pressure, possibly due to a better contact. Meanwhile, further increasing the pressure introduces current leakage, which is detrimental for photo-carrier transportation. Besides, the PCE of large-scale devices with an active area of 1.02 cm2 has been provided in Figure S5 and Table S6. The hot-pressing-treated device shows a PCE of 6.10%, which is nearly 30% higher than that of the control process.

Figure 4

J–V characterization of the large scale (0.5 cm2) PSCs prepared with (a) pressing process and (b) hot-pressing process. The J–V curves are measured under AM 1.5G one sun illumination (100 mW/cm2) with a scan rate of 0.1 V/s.

To offer a better insight into the performance improvement caused by hot-pressing, the interfacial charge transfer and recombination dynamics are investigated. Figure a records the steady-state photoluminescence (PL) spectrum of the neat CsPbBr3, CsPbBr3/carbon, and hot-pressing CsPbBr3/carbon films. The PL spectrum peaked at ∼530 nm, pointing to the formation of CsPbBr3. Compared with the neat perovskite films, the PL intensity of the carbon-coated perovskite films has been greatly reduced (Figure a). This is due to an effective carrier transportation from the perovskite layer up to the carbon-carrier collection layer. Besides, the lower the PL intensity, the more effective the carrier transportation.[4,44] The hot-pressing process quenched the PL intensity to less than 1/2 of the initial intensity (perovskite/carbon film), supporting a better contact at the perovskite/carbon interface after hot-pressing. Meanwhile, for the time-resolved photoluminescence (TRPL) characterization, Figure b records the TRPL spectroscopy of the prepared films, the curves are fitted with a biexponential decay model of I = Ae – (t – t0)/τ1 + Be – (t – t0)/τ2, and the carrier lifetime is extracted as summarized in Table S7. Because a carrier-selective layer has been introduced, the faster component (τ1) can be assigned to the photo-carrier transportation, and the long lifetime τ2 can be considered the intrinsic radiative recombination. The average carrier lifetime (τave), which can be calculated by deriving the formula of τave = (a1τ12 + a2τ22)/(a1τ1 + a2τ2), has been calculated to evaluate the charge extraction ability. The τave of the CsPbBr3/FTO substrate without the carbon layer is approximately 7.61 ns, whereas the introduction of the carbon layer contributes to a reduced τave, and the hot-pressing device shows a faster PL decay, indicating an enhanced hole extraction.

Figure 5

(a) Steady-state PL spectrum of the perovskite films. The measurements are conducted under a beam of 400 nm laser illuminated from the FTO side. (b) TRPL decay curves of the perovskite films deposited on the FTO substrate. A 405 nm laser is illuminated from the FTO side. IRF refers to the instrument response function. (c) Dark J–V curves of the symmetrical FTO/perovskite/carbon structure. The J–V curves are measured under dark conditions with a scan rate of 0.1 V/s. (d) Nyquist plots of the as-prepared perovskite solar cells in the dark at 0 V.

Furthermore, the J–V characterization of various PSCs with a symmetric structure of FTO/CsPbBr3/carbon is tested under the dark condition (Figure c). A relatively lower dark current density has been identified in the hot-pressing device, which supports a reduction in charge recombination.[30,32,42] Apparently, the reduced carrier recombination is mainly due a better contact of the perovskite/carbon interface. Next comes to the electrochemical impedance spectra (EIS) characterization. In the dark with an applied bias of 0 V, the arc on the EIS spectrum can be assigned to the recombination resistance (Rr) of the device.[26] Here, Rr has been increased from 1.786 MΩ for the control device to 3.228 MΩ of the hot-pressing device (Figure d), further supporting a less possibility of carrier recombination in the hot-pressing device. Finally, the hot-pressing devices retain over 95% of their initial PCE after a 30 day-aging test (Figure S6), compared with 92% for the control device; thus, a better stability can be guaranteed.

Conclusions

To sum up, high performance and large-scale CsPbBr3 PSCs are prepared via introducing a simple hot-pressing process. The hot-pressing process improves the contact at the perovskite/carbon interface and contributes to a better carrier transportation. Via adjusting the applied pressure, a superior PCE of 8.34% has been achieved for the hot-pressing-assisted device in comparison with 7.40% (pristine) and 7.95% (pressing devices). More importantly, the PCE of large-scale devices received a ∼30% PCE improvement (from 5.1% to 6.9%) with the hot-pressing treatment, indicating a more effective effect for large area devices. Furthermore, the hot-pressing devices showed an enhanced reproducibility and stability. Overall, the hot-pressing method shows broad prospects for carbon-based PSCs, especially for large-scale PSCs.

Experimental Section

Materials

The SnO2 colloid precursor was purchased from Alfa Aesar (tin(IV) oxide). N,N-Dimethylformamide (DMF) was purchased from J&K Chemicals. Lead bromide (PbBr2, 99.0%) and cesium bromide (CsBr, 99.9%) were obtained from Xi’an Polymer Light Technology Corp. The commercial carbon paste (including graphene and carbon black) was purchased from Shanghai Mater Win New Materials Co., Ltd. The other materials if not stated were purchased from TCI. All materials were used without further purification.

Device Fabrication

All the devices were prepared in the atmosphere environment. Fluorine-doped tin oxide (FTO) glasses were washed with detergent, deionized water, acetone, and isopropyl alcohol (IPA) in sequence for 15 min. The as-cleaned FTO was treated with UV-Ozone before use. As previously reported,45 the Li-doped SnO2 precursor solution was obtained by mixing SnO2 (450 μL) colloid dispersion and LiCl (300 μL) aqueous solution (17 mg/4 mL) with water (2 mL). A thin SnO2 layer was spin-coated onto the as-cleaned FTO at a speed of 4000 rpm for 30 s and annealed at 160 °C for 30 min. The CsPbBr3 films were prepared with a modified method based on the multi-step spin-coating method reported by Tang et al.[40] In detail, the PbBr2/DMF solution (1.0 M) was stirred at 70 °C for 30 min before being spin-coated onto the SnO2 layer at a speed of 2000 rpm for 30 s. The substrates were preheated at 90 °C prior to spin coating. The prepared PbBr2 films were annealed at 90 °C for 40 min. After cooling down, 0.07 M CsBr methanol solution was spin-coated onto the PbBr2 film at 2000 rpm for 30 s and heated at 250 °C for 5 min. This process was repeated several times to fabricate the cesium lead bromide films. The obtained perovskite films were soaked in isopropanol for 30 min and dried at 250 °C again for 5 min. Finally, a carbon back-electrode was deposited on the perovskite film by the doctor-blade coating method. For the pressing device, the as-prepared device was placed glass face up on a flat table, with a flexible polytetrafluoroethylene (PTFE) thin film sandwiched between the carbon film and table as a buffer layer. Then, weights were added to the top of the glass to apply pressure. For the fabrication of hot-pressing PSCs, the coated carbon paste was heated at 100 °C for 5 min.

Characterizations and Measurements

The XRD patterns of the samples were measured on an Ultima IV X-ray diffractometer with Cu Kα radiation. The J–V curves and EIS data were collected with an electrochemical workstation (CHI-760). The morphologies of the perovskite films were recorded on a JSM-6510A SEM instrument. The UV–vis absorption spectra were measured with a Hitachi U-3900 instrument. Steady-state PL (excitation at 400 nm) was measured with an Edinburgh Instruments Ltd. instrument (FLS980). TRPL was measured on an FL980 with the time-correlated single photon counting method by excitation with a 405 nm laser.