3.1. Characterization of ZnCo2O4 NPs. Crystallographic information of the prepared ZnCo2O4 NPs was acquired and the corresponding pattern is shown in Figure 1(a). The diffraction signals of ZnCo2O4 are found at 2θ = 31.06˚, 36.7˚, 38.36˚, 44.72˚, 55.52˚, 59.1˚, and 64.96˚, corresponding to the (220), (311), (222), (400), (422), (511), and (440) planes, respectively [32, 33]. According to the XRD pattern, the prepared ZnCo2O4 is well consistent with the spinel phase. Figure 1(b) displays the TEM image of ZnCo2O4 NPs. These particles tend to aggregate with an average diameter of 20 nm.
The residual NH3 molecules on the surface of ZnCo2O4 may deteriorate its electrical properties and thus should be removed. The FT-IR experiment was adopted to detect the removal of NH3, and the corresponding infrared spectra before and after calcination are depicted in Figure 2. Before calcination, the characteristic stretching bands of NH3 molecules were observed at 3655–2597, 1753, and 826 cm−1, which are assigned to the N–H stretching mode, H–N–H bending vibration, and H–N–H rocking mode, respectively [34]. A significant absorption band was found at 1317 cm-1, which was attributed to NO3 groups from starting materials [35]. In addition, the two IR absorption peaks for the Zn–O and Co–O bonds were revealed at 685 and 561 cm-1, respectively [36]. After calcination, it is clearly seen that the absorption bands at 3655–2597, 1753, and 826 cm-1 were vanished, indicating that NH3 molecules were removed. The NO3 absorption signal was also greatly diminished and a trace was found at 1384 cm-1. The Zn–O and Co–O bonds still existed at similar positions. The results proved that NH3 molecules can be easily removed during annealing to further improve electrical properties of ZnCo2O4 NPs.
To identify the Zn:Co ratio in our prepared ZnCo2O4 NPs, the XPS measurements were carried out. Figure 3(a) shows the Co 2p band of ZnCo2O4, and the multicomponent band can be deconvoluted into four different states at 779.5 (2p3/2), 794.7 (2p1/2) for Co3+, and 780.6 (2p3/2), 795.7 eV (2p1/2) for Co2+, and two shake-up satellite peaks at 789.8 eV near Co 2p3/2 band and 804.9 eV near Co 2p1/2 band. The locations of these states are in good accordance with the previous literature [37, 38]. The Zn 2p band of the spinel ZnCo2O4 NPs is depicted in Figure 3(b), revealing two XPS peaks at 1021 (2p3/2) and 1044 eV (2p1/2) for Zn2+ [38]. The Zn:Co atomic ratio is calculated to be 1:2.19 based on the XPS band area, which is close to the designed ratio of ZnCo2O4 (Zn:Co = 1:2). H.Y. Chen and his coworker claimed that some Co3+ can occupy Co2+ or Zn2+ sites in the structure because of the similar ionic radii of Co and Zn, thus giving rise to the antisite defects (ZnCo) [39], which is energetically favored for p-type conductivity. The prepared ZnCo2O4 NPs in this study is expected to show similar feature that is beneficial for carrier transport in optoelectronic devices. The O 1s spectrum of the obtained spinel ZnCo2O4 NPs is shown in Figure 3(c). The main signal due to lattice oxygen (O2–) is observed at 529.4 eV that is in agreement with the previous report [40]. Besides, shoulder signals at a higher binding energy of 530.9 eV and 532.3 eV come from surface hydroxyl groups and chemisorbed oxygen [41].
The energy levels of ZnCo2O4 NPs were calculated from their UPS spectra, as shown in Figure 4. The work function (φw) is derived by subtracting the binding energy cutoff in the high binding energy region (around 16.71 eV) from He I photon energy (21.22 eV). Since the φw is defined as the energy difference between the Fermi level (EF) and the vacuum level (0 eV), the EF value of ZnCo2O4 NPs is determined to be -4.51 eV from Figure 4(a). Furthermore, the binding energy cutoff in the low binding energy region reveals the energy difference between the EF and the valence band (VB) level [42]. The low energy binding cutoff of ZnCo2O4 NPs is found at around 0.6 eV in Figure 4(b), indicative of its VB level at -5.08 eV. Compared with the PEDOT:PSS film (VB level = -5.02 eV) [43], the downshifted VB level of ZnCo2O4 NPs is matched better with the perovskite absorbing layer, which can improve the hole extraction from the perovskite to ZnCo2O4 HTL.
3.2. Morphological observation of the ZnCo2O4 and perovskite layers. The top-view SEM images of PEDOT:PSS or ZnCo2O4 NPs deposited on the FTO substrates are shown in Figures 5(a) and (b), respectively. PEDOT:PSS is a transparent polymer and hence the grains of low-lying FTO are clearly seen. Besides, many small cracks exist on the surface of PEDOT:PSS. In Figure 5(b), ZnCo2O4 NPs are homogeneously deposited on the FTO surface and the grains of FTO are not observable. The surface roughness of the ZnCo2O4/FTO substrate may become lower since the grains of FTO are completely covered by ZnCo2O4 NPs, as compared with the PEDOT:PSS/FTO substrate. To verify this, AFM technique was adopted to investigate the morphology and average roughness (Ra) of the prepared samples. Figures 5(c) and (d) show the topographic AFM images of PEDOT:PSS and ZnCo2O4 NPs on the FTO substrates, respectively, revealing similar morphological features to those of the top-view SEM images. Furthermore, the Ra values of PEDOT:PSS or ZnCo2O4 NPs deposited on the FTO are estimated to be 15.1 and 6.65 nm, respectively. The result reveals that ZnCo2O4 NPs can serve as a better surface modifier for FTO substrates than PEDOT:PSS, which is beneficial for improving interfacial contact and hole extraction between ZnCo2O4 NPs and the perovskite [44]. The cross-sectional SEM images of ZnCo2O4 NPs layer and PEDOT:PSS film can be seen in Figure S1(a) and (b) in the Supplementary Information, the thickness of ZnCo2O4 NPs layer and PEDOT:PSS film were estimated to be ca. 65 nm and ca. 40 nm, respectively.
Figures 6(a) and (b) show the top-view SEM images of the perovskite deposited on PEDOT:PSS or ZnCo2O4 NPs, respectively. No pinholes could be found for both perovskite films. The grain size of perovskite crystals on PEDOT:PSS is estimated to be in the range of 100–180 nm, while larger perovskite crystals with grain sizes of 200–300 nm were observed on ZnCo2O4 NPs, as shown in Figure 6(b). As mentioned in the previous part, the lower surface roughness of the ZnCo2O4 layer helps to form larger sizes of perovskite grains, as compared with PEDOT:PSS film [45]. The high-quality perovskite film grown on ZnCo2O4 NPs is expected to exhibit higher photocurrent and conversion efficiency of PSCs.
3.3. Electrical Investigation of ZnCo2O4 NPs and PEDOT:PSS film. To investigate the hole transport ability of ZnCo2O4 NPs and PEDOT:PSS film, hole-only devices with the structure of FTO/ZnCo2O4 NPs or PEDOT:PSS/Ag were fabricated and evaluated. The electron-only device with the configuration of FTO/TBABF4-doped PC61BM/PEI/Ag was also fabricated for comparison. The corresponding current-voltage characteristics of the three devices are depicted in Figure 7, indicating that the ZnCo2O4 NPs device exhibits higher current and better hole transport capability than PEDOT:PSS film. Figure S2 in the Supplementary Information displays hole mobility (μh) of ZnCo2O4 NPs and PEDOT:PSS film, which is inferred from the space-charge limited current equation J = (9/8)εε0μh(V2/L3). The μh values of ZnCo2O4 NPs layer and PEDOT:PSS film are calculated to be 9.14×10-2 and 8.52×10-5 cm2/Vs, respectively. The obtained μh of PEDOT:PSS film is close to the reported value in the literature [46]. It is seen that our ZnCo2O4 NPs layer has a hole mobility by 3 orders of magnitude higher than that of the PEDOT:PSS film. Moreover, we found that the device FTO/TBABF4-doped PC61BM/PEI/Ag shows similar current-voltage behavior to the one based on ZnCo2O4 NPs, implying equivalent carrier transport capabilities of holes and electrons in our final inverted device architecture of FTO/ZnCo2O4 NPs/perovskite/TBABF4-doped PC61BM/PEI/Ag. The balanced carrier transport also helps to reduce the hysteresis effect of devices.
3.4. Optical Investigation of ZnCo2O4 NPs and Perovskite Layers. Figure S3(a) in the Supplementary Information shows the transmission spectra of the ZnCo2O4 NPs layer and PEDOT:PSS film from 315 to 750 nm. The transmittance was measured to be 55–90% in the range of 375–650 nm and even higher over 90% in the rage of 650–750 nm for both samples with similar spectral shapes. Therefore, we speculate that the amount of incident photons entering into devices is close. The absorption spectrum of the ZnCo2O4 NPs layer is shown in Figure S3(b) and its optical bandgap (Eg) of 3.7 eV was estimated from the absorption edge around 335 nm. From UPS and absorption measurements, the conduction band (CB) level of ZnCo2O4 NPs is determined to be -1.38 eV, while the lowest-unoccupied molecular orbital (LUMO) of PEDOT:PSS is referred to the previous literature (LUMO = -3.4 eV) [47]. The relatively high CB level of ZnCo2O4 NPs can reduce electron transport from the perovskite to FTO and carrier recombination inside devices.
The steady-state PL spectra of the perovskite on the FTO substrate, PEDOT:PSS film, and ZnCo2O4 NPs layer are indicated in Figure 8(a). It is clearly seen that the perovskite deposited on the FTO substrate has the highest PL intensity, while the one on the ZnCo2O4 NPs layer owns the lowest PL emission. The reduced PL emission implies hindrance of electron−hole pair recombination and improvement of JSC and FF of PSCs [27,48]. Furthermore, the TR-PL decay experiment was performed and the obtained PL decay curves of the perovskite on FTO, PEDOT:PSS film, and ZnCo2O4 NPs layer are shown in Figure 8(b). The PL decay curves agree well with a biexponential decay fitting and corresponding lifetimes of τ1, τ2, and τavg are listed in Table S1 in the Supplementary Information. It is reported that fast decay (τ1) originates from nonradiative capture of free carriers and the slow decay (τ2) comes from radiative recombination of remaining excitons [27]. The τavg is determined by the equation τavg =Σi(Aiτi2)∕Σi(Aiτi), where Ai values is derived from the fitted curve data [49]. Generally, the shorter carrier lifetime indicates more efficient charge extraction. The τavg value of the perovskite on FTO was calculated to be 107.17 ns, and it decreased to 88.61 and 39.98 ns when the perovskite was deposited on the PEDOT:PSS film and ZnCo2O4 NPs layer, respectively. This result indicates more effective charge extraction by the ZnCo2O4 NPs layer from the perovskite active layer as compared with the PEDOT:PSS film.
3.5. Device Evaluation. The p-i-n device structure of the inverted PSC based on ZnCo2O4 NPs HTL is shown in Figure 9(a), revealing a sandwiched architecture of FTO/ZnCo2O4 NPs/Cs0.05FA0.8MA0.15Pb(Br0.15I0.85)3/TBABF4-doped PC61BM/PEI/Ag. Figure 9(b) shows the cross-sectional SEM micrograph of the whole device, revealing the thickness of FTO, ZnCo2O4 NPs layer, perovskite, PC61BM+PEI, and Ag electrode to be 500, 60, 550, 35, and 135 nm, respectively. The energy level diagram of the whole device is illustrated in Figure 9(c). The VB and CB levels of ZnCo2O4 NPs have been discussed in the previous part, while the energy levels of the other components were referred to the previous reports [43, 50, 51]. In our device architecture, electrons can be successfully extracted from the perovskite absorber and transport to the Ag electrode through PC61BM+PEI, while holes migrate gradually from the perovskite layer through ZnCo2O4 NPs and are collected on the FTO electrode. The J-V curves of the devices measured under AM 1.5 G are shown in Figure 9(d), and the measured parameters including JSC, VOC, FF, PCE, and series resistance (RS) are summarized in Table 1. The optimized device based on ZnCo2O4 NPs showed a VOC of 0.92 V, a JSC of 19.85 mA/cm2, a FF of 67.19%, and a PCE of 12.31% in the reverse scan, which is significantly higher than the one based on PEDOT:PSS (VOC = 0.79 V, JSC = 17.23 mA/cm2, FF = 59.77%, and PCE = 8.11%). The statistical distribution of 20 individual devices for all photovoltaic parameters is depicted in Figure S4 in the Supplementary Information. It can be seen that our devices possessed good reproducibility and PSCs based on ZnCo2O4 NPs showed relatively higher photovoltaic parameters. The improved device performance is mainly ascribed to the increased JSC value and energy level matching between ZnCo2O4 NPs/perovskite interface. Hysteresis index (HI) can be used to describe the hysteresis behavior of PSCs according to the equation HI = (PCEreverse – PCEforward)/PCEreverse [52]. The PSC based on ZnCo2O4 NPs has a smaller HI value of 0.043 as compared with that based on PEDOT:PSS (HI = 0.36). As a result, the reduced hysteresis of the PSC based on ZnCo2O4 NPs is in accordance with electrical measurements in the previous part. The normalized PCE evolution of the PSCs based on ZnCo2O4 NPs and PEDOT:PSS is shown in Figure 9(e) for comparison. The PSC based on ZnCo2O4 HTL retained 85% of its initial efficiency after 240 hours storage under a halogen lamps matrix exposure at room temperature, whereas the PCE of the device based on PEDOT:PSS HTL dropped to only 0.5% of its initial efficiency after 144 hours storage. Such fast deterioration can be attributed to the acidic nature of PEDOT:PSS causing corrosion to the perovskite and FTO substrate. Therefore, the use of inorganic ZnCo2O4 HTL is highly beneficial for the device stability. As mentioned in the Introduction, the device using CuCo2O4 as the HTL retained 71% of initial PCE after 96 hours storage under a continuous yellow light irradiation.27 Our result reveals that ZnCo2O4 is a better candidate for the fabrication of stable PSCs. Figure 9(f) shows the EQE spectra and integrated current density of devices as a function of wavelength using ZnCo2O4 NPs and PEDOT:PSS as the HTL. The results demonstrate that the device based on ZnCo2O4 NPs has a higher photon-to-electron conversion capability from 300 to 750 nm compared to that based on PEDOT:PSS. The integrated current density for the devices based on ZnCo2O4 NPs and PEDOT:PSS was calculated to be 18.4 and 15.45 mA/cm2, respectively, which are similar to the JSC values in Table 1.