1 Introduction
Perovskite solar cells (PSCs) have emerged as a pioneer in the area of future solar cells owing to their outstanding photovoltaic performance and low cost [1–5]. However, the instability would be challenging for long-term stability due to low tolerance to temperature, humidity and residual solvents [6,7]. Recently, inorganic perovskites have attracted great attention because of their thermal stability and enhanced moisture resistant compared to hybrid counterparts [8–11]. The resultant inorganic CsPbX3 halides are featured with high charge carrier mobility, remarkable light absorption and considerable stability against humidity and heat [12–16].
To increase power conversion efficiency (PCE) of the CsPbBr3 device, a challenging and promising route is to optimize the crystallization process and adjust interfacial energy-level differences to improve charge extraction [17,18]. Since graphene-based derivatives possess excellent electronic properties, they have been investigated as candidates for doping materials and interfacial materials. Functional additives could tailor grain boundary trap states to annihilate defect states [19]. For example, polyaniline/graphite composites incorporated into carbon electrode can not only regulate work function, but also enhance back electrode’s ability to extract holes [20]. Kırbıyık et al. [21] introduced nitrogen doped carbon quantum dots into organic perovskite precursor to decrease voids or defects in perovskite domains for controlling the crystallization process to prevent the aggregation of perovskite. Yuan et al. [22] presented an enhanced PCE of 7.93% by inserting quantum dots between the perovskite absorption layer and transport layer. However, the cumbersome preparation and poor repeatability for devices is inevitable, which is an obstacle for practical applications.
Herein, we optimized a strategy to use GOQDs with abundant C and O active sites for participating in the crystallization of perovskite film to improve electron extraction performances and effectively reduce the possibility of deep charge traps forming at grain boundaries. The CsPbBr3 PSCs with GOQDs demonstrated better performances than other devices reported in literature. The performance comparison of CsPbBr3 PSCs with other graphene materials is shown in Table S1 (cf. Electronic Supplementary Material, ESM). The GOQDs were prepared by exfoliation method and dispersed evenly in N,N-dimethylformamide, which was deposited at the interface of TiO2/CsPbBr3 to optimize crystal size and accelerate electron extraction. We fabricated all-inorganic PSCs with a structure of fluorine-doped tin oxide (FTO)/TiO2/GOQDs@CsPbBr3/carbon without hole transport layer (HTL). This surface modification method could achieve a high PCE up to 9.16%, which is easier to operate and has excellent repeatability. These findings indicate that graphene-based derivatives can not only be used as interface modification materials, but can also be partially doped into the crystalline layer, affecting crystalline morphology and passivation defects for high-performance PSCs.
2 Experimental
2.1 Materials
CsBr (≥ 99.99%) and PbBr2 (≥ 99.99%) were purchased from Xi’an p-OLED. N,N-dimethylformamide (DMF, ≥ 99.9%), TiCl4 (≥ 99.5%), and ethanol were bought from Aladdin. TiO2 paste (18 NR) was purchased from Dyesol (Australia). All chemicals were used without any further purification.
2.2 Synthesis of GOQDs
The GOQDs were synthesized via hydrothermal treatment of graphene oxide material, which was prepared following Hummers’ method reported in our previous study [23,24]. First, 100 mg graphene oxide was ground for half an hour and then transferred to 10 mL of DMF for 60 min ultrasonic dispersion to obtain a homogeneous solution, and hydrothermal reaction was carried out at 200 °C for 5 h and cooling to room temperature naturally. The product was filtered by 0.02 µm microporous membrane and a brown solution was obtained, the black precipitate was discarded.
2.3 Fabrication of perovskite solar cell devices
All fabrication processes were performed under air atmosphere. The FTO glass substrates were etched for desirable patterns and purified by distilled water, acetone, 2-propanol, and ethanol, respectively. The compact TiO2 layer and mesoporous TiO2 layer were spin-coated onto FTO substrate, respectively [25]. Then, the TiO2 layers were treated in TiCl4 aqueous solution at 75 °C for 30 min, and annealed at 450 °C for 30 min in air. The GOQDs in DMF solution was deposited onto the TiO2 layer at 2500 r·min−1 for 30 s, followed by a heating at 90 °C for 30 min. This step was repeated for three times to get more GOQDs between TiO2 and perovskite film. The perovskite film was acquired by using multistep solution-processed method. PbBr2 in DMF solution was covered on the above surface at 3500 r·min−1 for 30 s and annealed at 90 °C for 30 min. Then, methanol solution of CsBr was spin-coated on top of the dried PbBr2 at 2000 r·min−1 for 30 s and the film was heated at 250 °C for 5 min. Finally, a carbon back-electrode coated on the perovskite layer by blade coating with an active area of 0.09 cm2 and then heated at 120 °C for 20 min.
2.4 Characterizations
X-ray diffraction patterns were obtained using X-ray diffractometer (XRD, Rigaku, D/MAX-2500/PC). The morphology of the perovskite films was characterized by scanning electron microscopy (SEM, JEOL-JSM 6700F). The GOQDs were characterized by transmission electron microscopy (TEM, JEM-2100F) and atomic force microscope (AFM, Multimode8, Bruker). The steady photoluminescence (PL) and time-resolved photoluminescence (TR-PL) were measured using the FLS1000. The photocurrent density–voltage (J–V) curves were measured by using an electrochemical workstation (CHI750e) under a solar simulator (Newport 96000, USA) with the irradiation of 100 mW∙cm–2 (AM 1.5). The dark J–V curves were measured at a scan rate of 10 mV∙s−1 (–1.6 to 1.6 V). The spectral response was obtained by an incident photon-to-current collection efficiency measurement system (EQE, Newport Power Meter 2936-C). The absorption spectra were analyzed by an ultraviolet-visible spectrophotometer (UV-3900, HITACHI) at room temperature. X-ray photoelectron spectroscopy (XPS) was performed (ESCALAB Xi+, ThermoFisher). In our work, all fabrication processes and measurements were carried out under ambient condition.
3 Results and discussion
The schematic illustration of constructing an all-inorganic PSC device is presented in . Both c-TiO2 and m-TiO2 layers were spin-coated onto the FTO substrate. To realize effective charge extraction, a GOQDs solution was dropped onto the dried TiO2 layer. The CsPbBr3 film was prepared by a solution-process method [26]. Finally, conductive carbon slurry was applied to cover solar cell by a doctor-blade method. The CsPbBr3 device with GOQDs consists of a layered structure of FTO/TiO2/GOQDs@CsPbBr3/carbon.
We used TEM, UV–Vis and AFM to analyze the structure, optical properties, and morphology of GOQDs (Fig.1). Fig.1(a) presents the morphology of GOQDs with a size of circa 4 ± 1 nm, and high resolution TEM image (Fig.1(b)) displays the interlayer spacing of 0.3 nm. There are characteristic GOQD peaks at the UV–Vis absorption spectrum (Fig.1(c)) because of edge effects and strong quantum confinement [27]. Meanwhile the GOQDs reveal great dispersion in methanol solution with a bright blue–green color under the illumination of 365 nm wavelength (inset of Fig.1(c)). AFM image (Fig.1(d)) shows that the synthesized GOQDs were well dispersed without apparent agglomeration and had a relatively uniform size distribution between 2 and 5 nm. The AFM image confirms that GOQD particles do not react easily with each other and can be stable in DMF at room temperature for two weeks.
The device structure, SEM cross section image, energy bandgap diagram, and photoelectronic properties are shown in Fig.2 and Tab.1. The devices were made of a stack of FTO/TiO2/GOQDs@CsPbBr3/carbon as presented in Fig.2(a). Fig.2(b) shows the cross-sectional SEM image of the device based on GOQDs/perovskite layers, which demonstrates that the mesoporous titanium dioxide layer and GOQDs have cross contact with the perovskite crystal layer and they are in a situation of mutual penetration. Compared to the pristine perovskite (Fig. S1, cf. ESM), the introduction of GOQDs can tailor the crystallization of CsPbBr3 and increase the size of perovskite CsPbBr3 in all dimensions. The relative energy levels of different layers are shown in Fig.2(c). Cyclic voltammetry was used to characterize energy level of GOQDs as described in the references [28,29]. This finding also illustrates the beneficial effect of GOQDs in promoting the transfer of electrons from the perovskite absorber to the current collector [30]. Fig.2(d) depicts the J–V curves of pristine PSC and PSC with GOQDs under AM 1.5G illumination, and Tab.1 shows their corresponding photovoltaic parameters. Without GOQDs, the PSC device exhibits a PCE of 5.43% with Voc = 1.26 V, Jsc = 6.23 mA∙cm–2, and FF = 69.2%. With the introduction of GOQDs to induce crystallization and modify the TiO2/CsPbBr3 interface, the PCE, Voc, Jsc and FF are improved to 9.16%, 1.38 V, 8.54 mA∙cm–2, and 77.6%, respectively. All the photovoltaic parameters are enhanced, which further confirms that the presence of GOQDs can lead to high crystallinity of CsPbBr3 and efficient electron extraction at the TiO2/CsPbBr3 interface [31]. The TRPL spectra (Fig. S2, cf. ESM) and related parameters (Table S2, cf. ESM) further confirm the reduced non-radiative recombination of photo-generated carriers in the perovskite film with GOQDs owing to the enhanced PL decay lifetime [32]. The charge carrier lifetime could be determined by fitting the decay curves by bi-exponential function. The average lifetimes were calculated to be around 1.79 and 4.29 ns for perovskite films with and without GOQDs, respectively. The decreased carrier lifetime can be attributed to faster charge carrier transfer process from the perovskite to the TiO2 layer, which is helpful for efficient charge transport and extraction. This is related to the improved interface contact properties due to enhanced conductivity and reduced defects or traps because of GOQD passivation. In addition, as shown in the Hall Effect parameter (Table S3, cf. ESM), the presence of GOQDs can increase carrier mobility.
Tab.1 Photovoltaic parameters of CsPbBr3 fabricated with or without GOQDs |
| PSC device | Jsc/(mA∙cm–2) | Voc/V | FF/% | PCE/% |
|---|---|---|---|---|
| Pristine PSC | 6.23 | 1.26 | 69.2 | 5.43 |
| PSC with GOQDs | 8.54 | 1.38 | 77.6 | 9.16 |
The XRD, XPS and SEM images of the films are demonstrated in Fig.3 to validate the existence of GOQDs. Fig.3(a) presents the XRD patterns of various perovskite films. For the CsPbBr3 films were prepared from fewer spin-coating cycles (< 6 cycle), three diffraction peaks appear at 11.7°, 29.4° and 33.3° corresponding to the PbBr2-riched phase of CsPb2Br5. The distinctive peaks at 15.5°, 21.6° and 30.7° correspond to the (100), (110) and (200) planes of CsPbBr3, respectively [33]. When repeated processes are up to 6 cycles, the intensities of the characteristic peaks at the two crystal planes (100) and (110) are higher than those under fewer spin-coating cycles and the CsPb2Br5-phase generation is inhibited, indicating a change from CsPb2Br5 to CsPbBr3 phase occurs with different numbers of CsBr spin coating, which implies that a high-purity CsPbBr3 layer is produced. Meanwhile, as shown in Fig. S3 (cf. ESM), the introduction of GOQDs does not change the main diffraction peaks of CsPbBr3 crystals, but the crystallinity can be improved with enhanced diffraction intensity of the (110) plane. In particular, the GOQDs-CsPbBr3 film achieves better degree of crystallinity because of increased diffraction intensity at 21.6°, and suppresses the growth of CsPb2Br5 crystals due to weaker diffraction peak at 29.4°. As a result, the films with added GOQDs have a higher crystallinity than the pristine CsPbBr3 film. XPS measurements (Fig.3(b)) indicate the contents of C and O elements in the FTO/TiO2/CsPbBr3 and FTO/TiO2/GOQDs@CsPbBr3 structure, which provides important evidence for the existence of GOQDs in the crystalline layer. In the top-view SEM images for the pristine perovskite film and the modified perovskite film, similar morphologies are observed (Fig.3(c) and 3(d)). Some voids were observed at the grain boundary in Fig.3(c), which is the main manifestation of the defect problem of the perovskite layer [19]. Fig.3(d) shows that the grains of the CsPbBr3 film with GOQDs are more unambiguous and legible than those of the pristine CsPbBr3 film. The average grain size of the CsPbBr3 film with GOQDs is circa 1.22 μm with much reduced grain boundaries. The increased grain size could enhance the crystallinity and restrain the structural defects associated with the pinholes. This also explains that GOQDs with plentiful carbon active sites participate in the crystallization of perovskite film. Electron mobility can be enhanced by increasing the crystallinity and purity of the perovskite film and reducing the intermediate phase, which improves solar cell device performances [34].
To investigate the effects of adding GOQDs on the photovoltaic performance of solar cells, we analyzed the bandgap and photoelectrical properties of the devices as presented in Fig.4. Photophysical characteristics of different CsPbBr3 films were examined. Fig.4(a) shows the UV–Vis absorption spectra of fabricated films, which intensity is highly related to the film thickness. The perovskite-structured CsPbBr3 film shows an absorption onset at 515 nm and weak absorption beyond 540 nm, which is mainly attributed to its large bandgap of 2.3 eV. The film modified by GOQDs does not change absorption property of the film and absorbs the light effectively that has not been fully absorbed by blank perovskite ranging from 540 to 800 nm. The illustrated curve (inset of Fig.4(a)) shows the relationship between photo energy (hv) and (ahv)2, which is obtained from the absorption spectrum. Based on the Kubelka–Munk equation, the optical bandgap (Eg) of the CsPbBr3 is 2.3 eV, which is compliance with those previously reported [8]. It is interesting that the PL peak intensity of TiO2 film was reduced by 24% with the introduction of GOQDs (Fig. S4, cf. ESM). This indicates that trap states were effectively passivated as a result of improved TiO2 film quality with the presence of GOQDs. Meanwhile, it indicates that efficient PL quenching occurred because of better charge transfer between electron transport layer (TiO2) and perovskite film [35]. According to the UV–Vis and UPS curves (Fig. S5, cf. ESM), the valence band maximum and conduction band minimum of the TiO2 electron transport layer with GOQDs are −7.2 and −3.92 eV, respectively. The incident photon-to-current efficiency (IPCE) and integrated current density as a function of wavelength are shown in Fig.4(b), which match well with the values from the J–V curve. Therefore, introducing GOQDs as an interfacial modification layer can suppress the recombination of interfacial radiation and enhance charge transfer capability.
For the CsPbBr3-based devices, the IPCE values were higher than those of the control PSC while all integrated Jsc was almost identical to the Jsc values extracted from the J–V measurements. Consistent power output curves were recorded at bias voltage at the maximum power point to verify the validity of the power output improvement, as shown in Fig.4(c). Continuous power output, output current density and PCE for 300 s or more are consistent to the corresponding values from the J–V measurement, which ensures good reliability of the PV performance analysis. To detail the charge transport mechanism within the device, we measured J–V data in the dark, as shown in Fig.4(d). A logarithmic plot of current density versus voltage (log J versus V) was investigated and as shown in the plot, the leakage current of the doped perovskite device is clearly lower, which is consistent with the previous J–V tests. According to the literature [36–38], the ideal factors nid of the devices with and without GOQDs are respectively 1.5 and 1.6 by using the slope of V and log J functions. The dark J–V curves reveal that non-zero slopes are an inevitable consequence of charge-carrier mobility. Low mobility represented by high slope causes an apparent shunt resistance (Rsh). Meanwhile, it appears that the GOQDs-added devices show smaller leakage current, suggesting higher shunt resistance (Rs) of the device.
The long-term stability of all-inorganic PSC devices with and without GOQDs were analyzed by the variation of J–V curves with the storing days (Fig.5). The devices were stored at ambient condition with relative humidity of about 35% without any encapsulation. The XRD patterns in Fig. S6 (cf. ESM) show that the CsPbBr3 film with GOQDs keeps high crystallinity after 8 d in atmosphere. There is no obvious crystal phase change after eight days of storage. The peaks at 30.4° and 27.8° correspond to CsPbBr3 and CsPb2Br5, and their diffraction intensities are slightly increased, which may be one of the reasons for the improved device performance [39]. After 40 d, the PCE of the devices without GOQDs decreases by 25% at ambient condition (Fig.5). In contrast, the devices with GOQDs maintain 94% of initial PCE under the same experimental conditions, reflecting a remarkable long-term stability. This increased stability derived from introducing GOQDs can be attributed to the protective effect and the surface defect.
Sixty-two devices were prepared and measured under the optimized conditions to study the reproducibility of the PSCs with GOQDs. The statistical data of the device performance parameters are shown in Fig.6. The averages of PCE, FF, Voc and Jsc are 8.9%, 75%, 1.38 V and 8.5 mA∙cm–2, respectively and close to the best device parameters, indicating the device with GOQDs has high reproducibility, which is mainly attributed to the high-quality perovskite film.
4 Conclusions
In this work, highly dispersible GOQDs nanoparticles were used as interface modification in CsPbBr3 perovskite. We presented the use of GOQDs at TiO2/CsPbBr3 interface as a valid pathway to accomplish highly efficient defect passivation carbon-based HTL-free CsPbBr3 PSCs, thereby achieving a prominent gain in device performance. GOQDs with compatible energy level regulates the crystallization and passivate the perovskite trap states, which is beneficial to collect electrons efficiently. Arising from promoting charge extraction, the GOQDs-treated PSCs without hole transport layer made by spin-coating obtained a higher output power conversion efficiency of 9.16% than 5.56% of the pristine device. This modified approach provides a valid and low-cost method for all-inorganic PSCs, opening up new pathway for future development of solar cells.