Understanding the Impact of Cu-In-Ga-S Nanoparticles Compactness on Holes Transfer of Perovskite Solar Cells

Although a compact holes-transport-layer (HTL) film has always been deemed mandatory for perovskite solar cells (PSCs), the impact their compactness on the device performance has rarely been studied in detail. In this work, based on a device structure of FTO/CIGS/perovskite/PCBM/ZrAcac/Ag, that effect was systematically investigated with respect to device performance along with photo-physics characterization tools. Depending on spin-coating speed, the grain size and coverage ratio of those CIGS films on FTO substrates can be tuned, and this can result in different hole transfer efficiencies at the anode interface. At a speed of 4000 r.p.m., the band level offset between the perovskite and CIGS modified FTO was reduced to a minimum of 0.02 eV, leading to the best device performance, with conversion efficiency of 15.16% and open-circuit voltage of 1.04 V, along with the suppression of hysteresis. We believe that the balance of grain size and coverage ratio of CIGS interlayers can be tuned to an optimal point in the competition between carrier transport and recombination at the interface based on the proposed mechanism. This paper definitely deepens our understanding of the hole transfer mechanism at the interface of PSC devices, and facilitates future design of high-performance devices.


Introduction
The p-i-n inverted planar organic-inorganic perovskite solar cell (PSC), a device whose structural concept originates from the donor-acceptor structure of organic solar cells (OSCs) [1,2], was first developed by Guo et al. in 2013 [3], and has attracted great attention from researchers owing to its transporting balance of both holes and electrons, its simple device structure and its facile fabrication process, in contrast to the alternatives [4][5][6][7][8]. The initial planar heterojunction of perovskite/C 60 is evolved into a core structure of perovskite/phenyl-C 61 -butyric acid methyl ester (PC 61 BM), which dominates almost all inverted planar PSCs structures, except for some inorganic components in place of PCBM [9][10][11]. At the same time, poly (3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS) in the device structure acts as a good hole-transport layer (HTL) because it has the highest occupied molecular orbital (HOMO), which matches well with that of perovskites [12].
Although PEDOT:PSS has been used successfully as HTL material in inverted planar perovskite solar cells, it still has some problems: firstly, it does not remain stable at high temperatures; second, its

Figure
1a,b shows cross-sectional SEM images of device structures of FTO/CIGS/perovskite/PCBM/ZrAcac/Ag without and with CIGS HTLs, respectively. The cross sections were prepared using the focus ion beam (FIB) tools. In both images, the MAPbI 3 perovskite layer has an average film thickness of approximately 360 nm, which ensures the effective absorption of incident solar spectrum. At the top of the perovskite layer, a continuous layer of PCBM covers the entire perovskite surface with a thickness fluctuating from 30 to 90 nm, which is due to the roughness of the perovskite surface derived from FTO substrates. Then, a thin layer of ZrAcac was spin coated onto the surface of the PCBM to tune the working function of the silver electrodes that are approaching the conduction band level of the PCBM [56]. A continuous layer of PCBM/ZrAcac is highly important in preventing the cross immigration of silver atoms and iodine ions [57]. In contrast to the structure without the CIGS interfacial layer, the anode interface with the CIGS modification becomes rougher, as shown in Figure 1b. Meanwhile, it is hard to tell that a compact and dense CIGS thin film covers the surface of the FTO substrates in Figure 1b, which is deemed to be a basic requirement for HTLs. Regardless, the perovskite film coated on the CIGS film looks compact ( Figure 1d).
(AXIS Ultra DLD, KRATOS Analytical Manchester, UK). The X-ray diffraction (XRD) patterns were recorded using a Bruker D8 X-ray diffractometer. An Atomic Force Microscope (AFM) (MFP-3D-Stand Alone, Asylum Research Abingdon-on-Thames, UK) was employed to investigate the film surface morphology.

Figure
1a,b shows cross-sectional SEM images of device structures of FTO/CIGS/perovskite/PCBM/ZrAcac/Ag without and with CIGS HTLs, respectively. The cross sections were prepared using the focus ion beam (FIB) tools. In both images, the MAPbI3 perovskite layer has an average film thickness of approximately 360 nm, which ensures the effective absorption of incident solar spectrum. At the top of the perovskite layer, a continuous layer of PCBM covers the entire perovskite surface with a thickness fluctuating from 30 to 90 nm, which is due to the roughness of the perovskite surface derived from FTO substrates. Then, a thin layer of ZrAcac was spin coated onto the surface of the PCBM to tune the working function of the silver electrodes that are approaching the conduction band level of the PCBM [56]. A continuous layer of PCBM/ZrAcac is highly important in preventing the cross immigration of silver atoms and iodine ions [57]. In contrast to the structure without the CIGS interfacial layer, the anode interface with the CIGS modification becomes rougher, as shown in Figure 1b. Meanwhile, it is hard to tell that a compact and dense CIGS thin film covers the surface of the FTO substrates in Figure 1b, which is deemed to be a basic requirement for HTLs. Regardless, the perovskite film coated on the CIGS film looks compact (Figure 1d). Thereafter, the device performances were investigated by tuning the coverage ratio and thickness of the CIGS films via spin coating speed ( Figure 2). Without CIGS, the J-V curve shows severe hysteresis, and the device achieves a conversion efficiency of 4.3%, VOC of 0.65 V, FF of 37.87% and JSC of 17.49 mA cm −2 in the backward scan. Without HTLs, the work function of FTO allows both holes and electrons to arrive at the perovskite/FTO interface without the blocking function of the electrons, leading to the intensive recombination of holes and electrons at the interface [38,39]. With the CIGS films, the open circuit voltages of the devices were improved remarkably to around 1 V, while also augmenting the fill factors ( Figure 2a). In general, the higher Thereafter, the device performances were investigated by tuning the coverage ratio and thickness of the CIGS films via spin coating speed ( Figure 2). Without CIGS, the J-V curve shows severe hysteresis, and the device achieves a conversion efficiency of 4.3%, V OC of 0.65 V, FF of 37.87% and J SC of 17.49 mA cm −2 in the backward scan. Without HTLs, the work function of FTO allows both holes and electrons to arrive at the perovskite/FTO interface without the blocking function of the electrons, leading to the intensive recombination of holes and electrons at the interface [38,39]. With the CIGS films, the open circuit voltages of the devices were improved remarkably to around 1 V, while also augmenting the fill factors ( Figure 2a). In general, the higher the spin speed, the thinner the CIGS film. With a spin-coating speed of 6000 r.p.m., the V OC reaches 1 V, along with the suppression of hysteresis, which reveals that a very thin layer of CIGS can effectively decrease the interface recombination to a large extent. As its thickness increases with the successive reduction of the spin-coating speed from 6000 r.p.m. to 4000 r.p.m. and 2000 r.p.m., the V OC still has a small augmentation, while the J SC is enhanced remarkably, to close to 21 mA cm −2 . Unexpectedly, the hysteresis becomes severe when the speed is decreased from 4000 to 2000 r.p.m., which demonstrates that some factors hinder the hole transfer efficiency and block the balance once again. This cannot be explained by the variation of CIGS thickness alone, and the relationship between the device performance and the speed dependent surface morphologies of CIGS interlayer is definitely nonlinear. Among all of the CIGS-modified devices, the most optimized device were recorded at the spin speed of 4000 r.p.m., and the best PCE was 15.16%, with a V oc of 1.04 V, J sc of 20.93 mA cm −2 , and FF of 69.77%. Figure 2b shows all of the EQE spectra, among which the device of 4000 r.p.m. shows the highest efficiency at over 80% in the visible light region. The integration of EQE spectra also confirms the current density tested by the simulator. The discrepancy between J-V and EQE is smaller than 5%, and belongs to the mismatch error [58]. the spin speed, the thinner the CIGS film. With a spin-coating speed of 6000 r.p.m., the VOC reaches 1 V, along with the suppression of hysteresis, which reveals that a very thin layer of CIGS can effectively decrease the interface recombination to a large extent. As its thickness increases with the successive reduction of the spin-coating speed from 6000 r.p.m. to 4000 r.p.m. and 2000 r.p.m., the VOC still has a small augmentation, while the JSC is enhanced remarkably, to close to 21 mA cm −2 . Unexpectedly, the hysteresis becomes severe when the speed is decreased from 4000 to 2000 r.p.m., which demonstrates that some factors hinder the hole transfer efficiency and block the balance once again. This cannot be explained by the variation of CIGS thickness alone, and the relationship between the device performance and the speed dependent surface morphologies of CIGS interlayer is definitely nonlinear. Among all of the CIGS-modified devices, the most optimized device were recorded at the spin speed of 4000 r.p.m., and the best PCE was 15.16%, with a Voc of 1.04 V, Jsc of 20.93 mA cm −2 , and FF of 69.77%. Figure 2b shows all of the EQE spectra, among which the device of 4000 r.p.m. shows the highest efficiency at over 80% in the visible light region. The integration of EQE spectra also confirms the current density tested by the simulator. The discrepancy between J-V and EQE is smaller than 5%, and belongs to the mismatch error [58]. In addition to the four representative kinds of devices, a series of devices with spin-coating speeds of the CIGS film ranging from 1000 r.p.m. to 6000 r.p.m. were investigated, and the results are summarized in Table S1, where the best values for each device parameter are listed in the forward and backward scans. For each kind of device, the performance data for a batch of 36 cells, measured in the same simulated solar spectra of AM 1.5G (100 mW cm −2 ), were collected and counted in the box chart in Figure 3. Compared to the device without CIGS, the average open circuit voltages of all the devices with different thicknesses of CIGS films were enhanced from around 0.5V to close to 1V, and show little difference, while they are obviously different with respect to JSC and FF. The large error bars for JSC and FF may be caused by the roughness of the FTO surface and the incomplete coverage of the CIGS nanoparticles. Therefore, it cannot be simply attributed to the thickness variation of the CIGS films. To figure out the mechanism behind this, the surface morphologies of all kinds of perovskite films just deposited on CIGS-coated FTO substrates were investigated ( Figure S1). All the perovskite films show compact crystalline grains ranging in size from 100 nm to 600 nm. In addition, CIGS thin films with different spin speeds had little impact on perovskite crystal formation. In contrast to the bare FTO substrates ( Figure S2b), the grain size of the perovskite film became smaller, clearly demonstrating the impact of the CIGS interfacial film. However, whatever the spin-coating speed of the deposition of CIGS, the variation in grain size and surface morphology of those perovskite films was really small. Hence, it is difficult to attribute the performance differences in the series of devices with CIGS to differences in their perovskite films. In addition to the four representative kinds of devices, a series of devices with spin-coating speeds of the CIGS film ranging from 1000 r.p.m. to 6000 r.p.m. were investigated, and the results are summarized in Table S1, where the best values for each device parameter are listed in the forward and backward scans. For each kind of device, the performance data for a batch of 36 cells, measured in the same simulated solar spectra of AM 1.5G (100 mW cm −2 ), were collected and counted in the box chart in Figure 3. Compared to the device without CIGS, the average open circuit voltages of all the devices with different thicknesses of CIGS films were enhanced from around 0.5V to close to 1V, and show little difference, while they are obviously different with respect to J SC and FF. The large error bars for J SC and FF may be caused by the roughness of the FTO surface and the incomplete coverage of the CIGS nanoparticles. Therefore, it cannot be simply attributed to the thickness variation of the CIGS films. To figure out the mechanism behind this, the surface morphologies of all kinds of perovskite films just deposited on CIGS-coated FTO substrates were investigated ( Figure S1). All the perovskite films show compact crystalline grains ranging in size from 100 nm to 600 nm. In addition, CIGS thin films with different spin speeds had little impact on perovskite crystal formation. In contrast to the bare FTO substrates ( Figure S2b), the grain size of the perovskite film became smaller, clearly demonstrating the impact of the CIGS interfacial film. However, whatever the spin-coating speed of the deposition of CIGS, the variation in grain size and surface morphology of those perovskite films was really small. Hence, it is difficult to attribute the performance differences in the series of devices with CIGS to differences in their perovskite films.  Eliminating the influence of the quality discrepancy of perovskite films themselves, the CIGS interfacial layer itself becomes the critical issue that needs to be examined. According to a previous report [40], Cu0.92In0.7Ga0.3S2 thin films were fabricated by spin coating the CIGS molecular precursor solution onto sterilized FTO glass, which was formed by dissolving CuCl, InCl3, GaCl3, and thiourea in DMSO. Figure 4a-f shows the surface morphologies of the CIGS thin films spin coated at different spin-coating speeds ranging from 1000 r.p.m. to 6000 r.p.m., respectively. These films are composed of individual CIGS nanoparticles which only fill in the valleys of FTO substrates ( Figure S2a), rather than fully covering the surface. As is generally known, the thickness of the thin film is related to the spin speed; that is, the higher the spin speed, the thinner the thin films and the smaller the nanoparticle size. The higher the spin speed, the smaller the nanoparticle size, but the higher the coverage ratio. Because the thickness of the CIGS will affect the transmittance, and the grain size will affect the roughness of the FTO surface, it will affect the performance of perovskite solar cells. In any case, the sharp corners of FTO grains were still exposed in all the samples. Eliminating the influence of the quality discrepancy of perovskite films themselves, the CIGS interfacial layer itself becomes the critical issue that needs to be examined. According to a previous report [40], Cu 0.92 In 0.7 Ga 0.3 S 2 thin films were fabricated by spin coating the CIGS molecular precursor solution onto sterilized FTO glass, which was formed by dissolving CuCl, InCl 3 , GaCl 3 , and thiourea in DMSO. Figure 4a-f shows the surface morphologies of the CIGS thin films spin coated at different spin-coating speeds ranging from 1000 r.p.m. to 6000 r.p.m., respectively. These films are composed of individual CIGS nanoparticles which only fill in the valleys of FTO substrates ( Figure S2a), rather than fully covering the surface. As is generally known, the thickness of the thin film is related to the spin speed; that is, the higher the spin speed, the thinner the thin films and the smaller the nanoparticle size. The higher the spin speed, the smaller the nanoparticle size, but the higher the coverage ratio. Because the thickness of the CIGS will affect the transmittance, and the grain size will affect the roughness of the FTO surface, it will affect the performance of perovskite solar cells. In any case, the sharp corners of FTO grains were still exposed in all the samples.
The composition of the CIGS nanoparticle films is characterized by Energy Dispersive Spectrum (EDS) mapping in Figure S3, where the elements Cu, In, Ga, and S are examined uniformly in CIGS nanoparticle film deposited on a bare glass substrate. The weak broad peak in the X-ray Diffraction (XRD) pattern of this CIGS nanoparticle film is indexed to the (112) plane of a CIGS chalcopyrite crystal structure ( Figure S4), which also implies that the film is very thin and not high in crystallinity due to its low annealing temperature [42,51]. However, the low-temperature solution process of our strategy affords much broader range of applications for CIGS films. As Figure 4a-f shows, the nanoparticle size of each film shrinks, as shown in the magnified images in each figure, while the coverage ratio increases with the increase in spin-coating speed. With respect to the device performance comparison mentioned above, both the grain size and the coverage of the CIGS interlayer films determine the charge transfer at the interface of perovskite/CIGS. There must be an optimal point between grain size and coverage ratio, where it has the best charge transfer efficiency. The charge transfer efficiency would more or less decline. In addition, the results of the Hall measurement confirmed the P type of the CIGS nanoparticle films, with an excellent mobility of 10.4 cm 2 V −1 S −1 , and a carrier concentration of 8.57 × 10 14 cm −3 , which is critical for efficient hole transport.  The composition of the CIGS nanoparticle films is characterized by Energy Dispersive Spectrum (EDS) mapping in Figure S3, where the elements Cu, In, Ga, and S are examined uniformly in CIGS nanoparticle film deposited on a bare glass substrate. The weak broad peak in the X-ray Diffraction (XRD) pattern of this CIGS nanoparticle film is indexed to the (112) plane of a CIGS chalcopyrite crystal structure ( Figure S4), which also implies that the film is very thin and not high in crystallinity due to its low annealing temperature [42,51]. However, the low-temperature solution process of our strategy affords much broader range of applications for CIGS films. As Figure 4a-f shows, the nanoparticle size of each film shrinks, as shown in the magnified images in each figure, while the coverage ratio increases with the increase in spin-coating speed. With respect to the device performance comparison mentioned above, both the grain size and the coverage of the CIGS interlayer films determine the charge transfer at the interface of perovskite/CIGS. There must be an optimal point between grain size and coverage ratio, where it has the best charge transfer efficiency. The charge transfer efficiency would more or less decline. In addition, the results of the Hall measurement confirmed the P type of the CIGS nanoparticle films, with an excellent mobility of 10.4 cm 2 V −1 S −1 , and a carrier concentration of 8.57 × 10 14 cm −3 , which is critical for efficient hole transport.
On the other hand, the CIGS film in a FTO/CIGS/Perovskite/PCBM/ZrAcac/Ag device should be thin enough for light transmission, because the optical absorption edge of CIGS is around 740 nm. Figure 5a presents the absorption spectra of the series of CIGS thin films. The film spin-coated at 1000 r.p.m. exhibits an obvious absorption edge at around 740 nm, which is the typical one of CIGS films [59,60]. With the increase in spin speed, the nanoparticles become smaller and the films become thinner, and the CIGS absorptions becomes consequentially weaker, which eliminates our worry that this window layer would absorb the sunlight in advance of the underlying perovskite active layer. When the spin speed was higher than that at 1000 r.p.m., the optical band gaps of these films became larger with small variations. Figure 5b shows us a direct view of the transmission of these CIGS films. The transmission of these CIGS films decreases with the decrease in spin coating speed. This is because the thicker CIGS films derived from a lower spin speed allow less light to be transmitted due to both absorption and reflection by CIGS-modified FTO glass, leading to lower current density in the devices (Figure 2). However, the most optimized cell was recorded with 4000 On the other hand, the CIGS film in a FTO/CIGS/Perovskite/PCBM/ZrAcac/Ag device should be thin enough for light transmission, because the optical absorption edge of CIGS is around 740 nm. Figure 5a presents the absorption spectra of the series of CIGS thin films. The film spin-coated at 1000 r.p.m. exhibits an obvious absorption edge at around 740 nm, which is the typical one of CIGS films [59,60]. With the increase in spin speed, the nanoparticles become smaller and the films become thinner, and the CIGS absorptions becomes consequentially weaker, which eliminates our worry that this window layer would absorb the sunlight in advance of the underlying perovskite active layer. When the spin speed was higher than that at 1000 r.p.m., the optical band gaps of these films became larger with small variations. Figure 5b shows us a direct view of the transmission of these CIGS films. The transmission of these CIGS films decreases with the decrease in spin coating speed. This is because the thicker CIGS films derived from a lower spin speed allow less light to be transmitted due to both absorption and reflection by CIGS-modified FTO glass, leading to lower current density in the devices ( Figure 2). However, the most optimized cell was recorded with 4000 r.p.m., rather than 5000 r.p.m. or 6000 r.p.m., which demonstrates that light transmission is not the only factor to determine the device current density and other performance parameters. Therefore, the band levels of our devices were investigated, as well as the hole transport ability of the CIGS interfacial layers, as described below.
Ultraviolet Photoelectron Spectroscopy (UPS) was applied to explore the band energy levels of these CIGS interfacial nanoparticles films ( Figure 6). The work function was calculated by the equation E F = 21.22 − E cutoff , and E F is the Fermi level, and E cuoff is the high binding energy cutoff. The work functions of CIGS-2000, CIGS-4000, and CIGS-6000 were 3.88 eV, 4.04 eV, and 4.10 eV, respectively. The valence band levels of the three typical kinds of CIGS-modified FTO substrates, along with bare FTO glass, were deduced from Figure 6a and 6b by the function E VB = E F + E 0, and E VB is VBM, E 0 is the low bounding energy tails [30,61]. The band gap of each corresponding kind can be calculated from Figure 6c. The band levels of the devices are collected in Figure 6d, which clearly depicts the difference in VBM mismatch between CIGS modified FTO substrates and perovskite films. The CIGS-4000 shows the best matching in VBM with that of the perovskite films, and the band level offset is the smallest (0.02 eV) among all the devices in this work. This accounts for the most efficient hole transfer at the anode interface. Meanwhile, the high CBM of each of the CIGS-modified FTOs also suppresses the electrons transferred from perovskite to FTO. This may be the main contribution to the augmentations of V OC and FF for the CIGS-based devices with respect to the device on a bare FTO substrate [39]. r.p.m., rather than 5000 r.p.m. or 6000 r.p.m., which demonstrates that light transmission is not the only factor to determine the device current density and other performance parameters. Therefore, the band levels of our devices were investigated, as well as the hole transport ability of the CIGS interfacial layers, as described below. Ultraviolet Photoelectron Spectroscopy (UPS) was applied to explore the band energy levels of these CIGS interfacial nanoparticles films ( Figure 6). The work function was calculated by the equation EF = 21.22 − Ecutoff, and EF is the Fermi level, and Ecuoff is the high binding energy cutoff. The work functions of CIGS-2000, CIGS-4000, and CIGS-6000 were 3.88 eV, 4.04 eV, and 4.10 eV, respectively. The valence band levels of the three typical kinds of CIGS-modified FTO substrates, along with bare FTO glass, were deduced from Figure 6a and 6b by the function EVB = EF + E0, and EVB is VBM, E0 is the low bounding energy tails [30,61]. The band gap of each corresponding kind can be calculated from Figure 6c. The band levels of the devices are collected in Figure 6d, which clearly depicts the difference in VBM mismatch between CIGS modified FTO substrates and perovskite films. The CIGS-4000 shows the best matching in VBM with that of the perovskite films, and the band level offset is the smallest (0.02 eV) among all the devices in this work. This accounts for the most efficient hole transfer at the anode interface. Meanwhile, the high CBM of each of the CIGS-modified FTOs also suppresses the electrons transferred from perovskite to FTO. This may be the main contribution to the augmentations of VOC and FF for the CIGS-based devices with respect to the device on a bare FTO substrate [39]. Based on all the understandings mentioned above, Figure 6e shows plausible carriers transport routes in our devices. Firstly, electrons and holes depart from photon generated excitons (Step 1). Holes will go through CIGS nanoparticles (Step 2) and finally arrive at FTO electrodes (Step 3), while electrons will be extracted to PCBM (Step 4). Since the CIGS layer is not a continuous and dense film with FTO grain corners exposed, it is probable for electrons to reach the surface of FTO and recombine with the holes accumulated there (Step 5). Hence, a competition definitely exists between Step 5 and Step 3. Here, the grain size and coverage ratio of those CIGS interlayers would play a role in this competition, in addition to the band alignment effect of CIGS modifications. For CIGS-1000 or CIGS-2000, the grain sizes are larger, but the coverage is smaller. The hole transfer ability of CIGS nanoparticles will be limited by inadequate transfer sites, and hence their hole transfer efficiency is not high enough, and the recombination of carriers at this interface must occur Based on all the understandings mentioned above, Figure 6e shows plausible carriers transport routes in our devices. Firstly, electrons and holes depart from photon generated excitons (Step 1). Holes will go through CIGS nanoparticles (Step 2) and finally arrive at FTO electrodes (Step 3), while electrons will be extracted to PCBM (Step 4). Since the CIGS layer is not a continuous and dense film with FTO grain corners exposed, it is probable for electrons to reach the surface of FTO and recombine with the holes accumulated there (Step 5). Hence, a competition definitely exists between Step 5 and Step 3. Here, the grain size and coverage ratio of those CIGS interlayers would play a role in this competition, in addition to the band alignment effect of CIGS modifications. For CIGS-1000 or CIGS-2000, the grain sizes are larger, but the coverage is smaller. The hole transfer ability of CIGS nanoparticles will be limited by inadequate transfer sites, and hence their hole transfer efficiency is not high enough, and the recombination of carriers at this interface must occur substantially. As the spin-coating speed increases, the grain size shrinks, while the coverage ratio increases. When the grain size is below some critical value, as shown in CIGS-5000 and CIGS-6000, the charge transfer function of CIGS nanoparticles will deteriorate remarkably. Regarding the tradeoff between the grain size and coverage ratio of CIGS interlayers, CIGS-4000 seems to be the optimal with respect to this competition. Moreover, the variation trend of the hysteresis in J-V curves also implies that smaller grain sizes result in more efficient hole transfer.
To further confirm the working mechanism of CIGS interfacial layers, photoluminescence (PL) and time-resolved photoluminescence (TRPL) were used to characterize the perovskite films deposited on the four typical kinds of substrates. As displayed in Figure 7a, compared with the device without CIGS, the PL intensity was quenched significantly by the CIGS films, indicating effective hole extraction of the CIGS film. Comparing CIGS-2000, CIGS-4000 and CIGS-6000, CIGS 4000 shows the highest quenching rate. This phenomenon further confirms the mechanism illustrated in Figure 5. This coincides with the data of TRPL (Figure 7b). The PL decay curves were fitted using the biexponential decay function. The charge transfer time includes tow processes; namely, the fast decay process (τ 1 ) and slow decay process (τ 2 ). In addition, the fast decay represents the charge extraction from the perovskite to the CIGS film; the slow decay represents the radiative recombination in the perovskite [24,62,63]. The detailed data are summarized in Table S2. Among them, CIGS-4000 acts as the fastest hole extraction layer, with 0.69 ns of τ 1 , while that of the bare FTO, CIGS-2000, CIGS-6000 are 11.67 ns, 3.37 ns, and 2.92 ns, respectively.   Figure 7c presents the Nyquist plot of the PSCs under dark at 0.7 V bias from 100 Hz to 1 MHz. As we know, the arcs at high frequency refer to the charge-transport resistance between the interface and perovskite, and those at low frequency are attributed to charge recombination [63][64][65]. Without CIGS modification, the smallest arc indicates rapid recombination of carriers at the interface between FTO and perovskite. Coinciding with the above results, with CIGS, the arc becomes larger, and therefore possesses slower carrier recombination, owing to the effective extraction by CIGS nanoparticles at the interface. Finally, the stability of the CIGS-4000 PSCs was checked. As for the  Figure 7c presents the Nyquist plot of the PSCs under dark at 0.7 V bias from 100 Hz to 1 MHz. As we know, the arcs at high frequency refer to the charge-transport resistance between the interface and perovskite, and those at low frequency are attributed to charge recombination [63][64][65]. Without CIGS modification, the smallest arc indicates rapid recombination of carriers at the interface between FTO and perovskite. Coinciding with the above results, with CIGS, the arc becomes larger, and therefore possesses slower carrier recombination, owing to the effective extraction by CIGS nanoparticles at the interface. Finally, the stability of the CIGS-4000 PSCs was checked. As for the chemical, oxygen and moisture stability in the air, the CIGS nanopartical layer can also protect the perovskite layer from being damaged. The device was tested at interval when stored in an inert environment without encapsulation. The parameters of J-V curves are shown in Figure 8a-b, and it retains almost 80% of its initial conversion efficiency value (Figure 8c). The FF degrades obviously, while the V oc and J sc are unexpectedly augmented. In any case, it exhibits good stability and great potential for future application.

Conclusion
In summary, the CIGS nanoparticles films were involved in evaluating the impact of the compactness of hole transport layers in the performance of PSCs. By changing the spin-coating speed from 1000 to 6000 r.p.m., the device at a spin-coating speed of 4000 r.p.m. achieved the optimal PCE of 15.16%, and Voc of 1.04 V. In the UPS results, CIGS-4000 had the smallest band level offset (0.02 eV) with the perovskite, and achieved the most efficient hole transfer at the anode interface, also demonstrating that it was optimal in terms of the competition between recombination and transport at the interface. The optimum was yielded by a synergetic effect of both grain size and coverage ratio of the CIGS interlayer covering the surface of FTO electrodes. PL quenching and TRPL also confirm the variation law, along with the results of their Nyquist plots. This paper answers the question as to the necessary compactness of hole transport layers that is sufficient to separate holes from photon-generated carriers, and also definitely deepens the scope of our understanding of the detailed function of carrier transport layers.

Conclusions
In summary, the CIGS nanoparticles films were involved in evaluating the impact of the compactness of hole transport layers in the performance of PSCs. By changing the spin-coating speed from 1000 to 6000 r.p.m., the device at a spin-coating speed of 4000 r.p.m. achieved the optimal PCE of 15.16%, and V oc of 1.04 V. In the UPS results, CIGS-4000 had the smallest band level offset (0.02 eV) with the perovskite, and achieved the most efficient hole transfer at the anode interface, also demonstrating that it was optimal in terms of the competition between recombination and transport at the interface. The optimum was yielded by a synergetic effect of both grain size and coverage ratio of the CIGS interlayer covering the surface of FTO electrodes. PL quenching and TRPL also confirm the variation law, along with the results of their Nyquist plots. This paper answers the question as to the necessary compactness of hole transport layers that is sufficient to separate holes from photon-generated carriers, and also definitely deepens the scope of our understanding of the detailed function of carrier transport layers.