Stable Perovskite Solar Cells Resist to Water without Encapsulation by P-Type Si Nws as Hole Collection Layers

INTRODUCTION
Perovskite solar cells only need 8 years to overcome barrier of dye sensitized solar cells (low efficiency) and its efficiency have increased from 3.8% in 2009 [1] to 22.7% in late 2017, [2] making this the fastest-advancing solar technology to date which its efficiency is comparable with silicon solar cells. That why perovskite solar cells became a hot topic. According to the Web of Science Database 2294 articles related to the perovskite solar cell were published in 2016 and 4353 articles have published from 2015 to April 2017. Among perovskite materials with formula of ABX3, the perovskite materials base on CH3NH3PbI3 (MAPbI3) has recently been identified as a promising absorber layer for solar cells because of high carrier mobility, long charge-carrier diffusion and high absorption coefficient [3]. Although perovskite solar cell overcomes efficiency limitation, it still suffers from instability.    
Long-term device stability in the air atmosphere is the most persistent issue that delays perovskite solar cell commercialization [4-7]. The origin of this instability is perovskite absorber materials which were degraded in presence of water, oxygen and ultraviolet light. To date, most reports characterize device stability in the absence of these extrinsic factors [4-7]. Significant attention has been brought towards improving the stability of the devices but it still has remained challenging [4-7]. 
Current state of the arte strategy for stabile perovskite solar cells can be classified as four ways. First, improving stability by replacing MA cation with other cations. For instance Gratzel and et al. Replace MA with Cs and Fa which lead to improve stability [8]. According to their report, FA0.9Cs0.1PbI3 film was stable for 19 h. In another report, Michael Saliba and Co-workers show that Cs x(MA0.17FA0.83)(1−x)Pb(I0.83Br0.17)3 film is stable for 250 h under operational conditions [9]. Second, adding additive in the perovskite solutions. Adding additive is another way to improve stability of perovskite solar cells. For example Zhao et al. fabricate perovskite solar cell with 300 h stability by adding polyethylene glycol (PEG) in the perovskite solution [10]. Grätzel and co-workers used phosphonic acid ammonium additives (4-ABPACl) forming the crosslinking of CH3NH3PbI3 perovskite via strong hydrogen bonding of the -PO(OH)2 and -NH3+ [11] which shows better efficiency and stability. It was stable at 45 °C for one week. Third, inorganic HTM: using inorganic HTM is another way to improve stability of perovskite solar cells.  Bian and co-workers introduced CuOx as HTL in the inverted planar p-i-n PVSCs,[12] in which the stability test of the unencapsulated device with CuOx flm suggested superior stability than that of device with PEDOT:PSS layer. In another research reported by J. You and et al. 1440 h stability with PCE of 16.1% was achieved by using NiOx as HTM [13]. In another report which was published in JACS, stability of 2h was achieved by using CuI as HTM [14]. Forth, encapsulation is last way to improve perovskite solar cells. For example Leijtens et al. demonstrated that the when CH3NH3PbI3−xClx perovskite devices with a mesoporous Al2O3 scaffold were encapsulated with an epoxy resin and a glass coverslip in a nitrogen-filled glove box was stable for 1000 h [15]. Yong and co-workers introduced a hydrophobic polymer Teflon as polymer encapsulation to enhance stability of perovskite solar cell under ambient atmosphere conditions and 30 d stability was achieved [16].  The PVSCs also showed stable behaviour over a certain time while it was even immersed in water.
To solve this problem we need to protect absorber layer from these extrinsic factors. Using suitable hole-transporting materials (HTMs) could be a bright idea. 
At present, state-of-the-art devices generally employ a p-type organic small-molecule or polymeric hole-conductor as hole-transporting layer (HTL), such as 2,2′,7,-7′-tetrakis(N,N-di-p-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD) and poly(3- hexylthiophene) (P3HT) to achieve the highest efficiencies in perovskite solar cells [17- 19]. As mentioned previously, a critical limitation on commercializing this technology comes from its instability that could be solved by introducing suitable HTMs that offer full performance without sacrificing long-term stability, scalability, and with low material [20].
Since HTL forms the outer layer in typical perovskite solar cells, it is conceivable that the HTL could protect perovskite structure by shielding it from atmospheric moisture [19, 21-23]. The amount of this protection will consequently depend on the different parameters such as permeability, hydrophobicity and density of the hole-transporting material [19, 24-26]. These additional properties of the HTL could have a strong impact on the overall stability of this type of solar cells. 
In the present work we introduce p-type Si Nano wires (Si NWS) as a hole transport material. As is well known, CH3NH3PbI3 are extremely sensitive to moisture in air but p-type Si Nano wires could protect it from moisture. In fact we observe very high stability when Si NWS were used as HTM. Beside this Si NWS has some other advantages, Si has excellent hole mobility (2.34 × 109 T-2.7 cm2/Vs) which is much higher than Li doped spiro-MeOTAD (2 × 10-2 cm2/ V s ), its hole mobility almost 1011 times higher than spiro as common HTM for perovskite solar cells which is remarkable [27]. However, Si NWS have higher hole mobility compare to the Si wafer [28, 29]. Using Si NWS have some advantages compare to the Si wafer such as wider band gap, higher hole mobility and higher surface area. Definitely increasing band gap is the most important benefit of Si NWS which bring it up as an ideal HTM. Quantum confinement could increase bandgap of silicone to even 3 ev [30]. Our results shows that using silicon wafer with band gap of 1.1 ev drop of the efficiency to 7% while using Si NWS with band gap of 1.8 ev boost efficiency to 23.8 %.  Currently, controlling environmental conditions to a moisture level of less than 1% is needed for device preparation because CH3NH3PbI3 films are not stable in a humid atmosphere but in presence of Si NWS controlling environmental conditions is not needed [19, 31-33]. To be stable in air without the use of any type of encapsulation is an important factor for the operation of any device which Si NWS as HTM provide it. We even tested its water resistance and it was quite stable (See video and its XRD in supporting information). This is excited results which encourage researcher to use this new HTM in their future works.

MATERIALS AND METHODS
Preparation of Si NWS
P-type Si nanowires were prepared by wet chemical etching of silicon wafers. P-type single crystalline silicon with orientation (100) from Siltronic was employed as raw wafer. For all measurements, the samples were cut into 2×2 cm2. Si wafers were cleaned by immersing in acetone, ethanol and HF (5 M) and DI water for 10 min respectively. The etching process was according to our previous work [34], briefly it was done in a sealed teflon vessel using a HF (5 M) and AgNO3 (0.02 M) solution at 50 ℃. 

CH3NH3PbI3 coating
In order to study the degradation of CH3NH3PbI3, CH3NH3PbI3 were coted on mesoporous TiO2 (m-TiO2) and Si NWs and their degradation were monitored by using XRD. In the case of TiO2, First FTO glass was cleaned according to our previous work [35]. Briefly, it was sonicated by using ethanol, acetone and dried by nitrogen gun.  Then compact layer of TiO2 was coated by spin coating. Typically 1 mL of tetraethyl orthotitanate (TEOT) was dissolved in mixture of 1mL of H2O, 1 mL HCl and 8 mL ethanol and was spin coated for 30 S with 2000 rpm. For coating m-TiO2, TiO2 paste purchased form Dyesol was diluted with ethanol (1:10 V:V) and spin coated for 45 S with 2000 rpm. CH3NH3PbI3 layer sequentially was deposited following procedures reported in paper 36. 0.4 M solution of MAPbI3 perovskite were prepared by dissolving PbI2 and CH3NH3I. It should be noted that all the solutions were then stirred for 1 h at 70 °C prior to use. The FTO/TiO2 substrates were next preheated at 180 ℃ and was removed from the hot plate, dipped into the solution and quickly pulled out. The perovskite crystallized on the heat substrate within 5 s as indicated by the colour change from yellow to dark black. Same procedure was used to deposited perovskite layer on the Si NWS. It should be notice that before coating perovskite layer on Si NWS, Si NWS was immersed in HF in order to remove SiO2 from its surface. 

Monitoring degradation of perovskite layer
 Degradation of perovskite layers were tested in air for 50 days by using XRD and Pl. Also we tested the stability of perovskite layer on Si NWS under direct flow of water. Characterization of Perovskite Films: Morphological characterizations were performed by scanning electron microscope (SEM). 

RESULTS AND DISCUSSION 
In order to study the stability of as fabricated perovskite layer XRD and Pl was used. Degradation of MAPbI3 on the m-TiO2 was studied as reference (PVSK-ref). XRD of PVSK-ref is shown in Fig. 1 a-c. Powder diagrams were recorded right after the dipping and after 1,2, and 7 days. Fig. 1a shows the XRD pattern of fresh PVSK-ref. Results show that PVSK-ref start to degrade after 1 day (Fig. 1 b) and it almost degraded in 7 days (Fig. 1 c).  
Thin films of MAPbI3 were grown on a hot Si (100) NWs which was briefly dipped into a hot mixture of MAI and PbI2. Then powder diagrams were recorded right after the dipping and after 1, 2, 3, 7 and 30 days. As seen, MAPbI3 on P-type Si NWS shows superior stability. Stability of MAPbI3 on Si NWS was studied for 50 days (PVSK-Si). Related XRD patterns are shown in Fig. 2 a-f.  We observed interesting and unusual behaviour in the case of formation of MAPbI3 on Si NWS.
Fig. 2 a shows the XRD pattern for fresh MAPbI3 on Si NWS. At first, deposited perovskite films do not show significant peak, but in the end it shows strong peak at 2θ=14 corresponding to (110). Its orientation has changed over the time. Fresh perovskite shows amorphous structures (Fig. 2 a). After 1days, MAPbI3 crystal start to form and orient along the (110) (Fig. 2 b). After 2 days, we observed unexpected results, crystal orient along the (202) (Fig. 2 c). Surprisingly we observe crystal orient again along (110) and (220) after 3 days (Fig. 2 d). After 7 days, crystal oriented mainly along (110) (Fig. 2 e). After 30 days, its crystal orientation almost was same as 3 days (Fig.2. f). It seems this is smart crystal and continual oriented in different way to be stable against moisture on so on. This was strange results that why we repeated it and got same results (see XRD for repeated sample in supporting information, Fig. S1). In addition, we tested stability of perovskite layer on Si NWs under direct flow of water. For this, above sample (PVSK-Si) that its stability was studied for 42 days was chosen. After 43 days, it was placed under direct flow of water (see video in supporting information). Surprisingly it quit stable after washing with water. Related XRD patterns are shown in Fig. 3.
Fig. 3a and b show its XRD pattern after 1 day of washing with water and 6 days after washing, respectively. As seen from the figure, it quite stable even after washing with water and after 50 days after preparation.  Briefly, since the first diagram shows a hardly crystalline simple, we may conclude that the system was in a quite undetermined state then and that the crystallites were rather small, especially since the sample underwent the cubic → tetragonal phase transition. The subsequent diagrams show the growth of the MAPbI3 phase; this growth manifested in varying ratios of the MAPbI3 lines. After about two days, an unknown phase appeared which could not be identified, at least not in the C, H, N, I, Pb system.
This phase later disappeared and so we can say that the stable target (CH3NH3)PbI3 was successfully produced. The stability of the presented perovskite could be compared with state of the art devices in term of stability (Table 1). Table 1 shows that all other reports without encapsulation were stable in humidly environmental while our device is stable under flow of water without any encapsulation.
Fig. 4a-b shows the PL results for TiO2/MAPbI3 film, PL intensity decrease over the time. We also use PL technic to demonstrate that Si NWS can extract holes. As seen from Fig. 4 c, PL of MAPbI3 on Si NWS has low intensity because hole extracted from perovskite layer.
 SEM and X-ray map were used to study the morphology and structure of Si NWS and perovskite layer. SEM results in Fig. 5 show that Si NWS have 60 nm diameters. MAPbI3 crystal has about 2 µm grain size. EDX elemental maps showing well-defined layers and good infiltration of lead and iodine inside Si NWS.
We propose three possible reasons for unusual stability of perovskite layer on Si NWS: first, hydrophobic properties of Si NWS. Si NWS has hydrophobic properties that repel water from its surface. Therefore water or moisture could not reach the surface of Si and therefore protect perovskite layer from water. Fig. 6 shows contact angel images for Si NWs. Its contact angel is about 130 degree which confirms its hydrophobic properties. Second, orientation of perovskite crystals changed over the time (this may happens to be stable against moisture). Third, as seen from the XRD results, perovskite crystals form very slowly on Si NWS (it forms after 1 day). We are thinking that slow formation of crystal lead to slow down the degradation rate of perovskite. We will discuss last two reasons in our future work.
In this work, SCAPS version 3.2.01(a Solar Cell Capacitance Simulator) software which is a one dimensional solar cell simulation program is used to simulate perovskite solar cells. This software is developed at Department of Electronics and Information Systems (ELIS) of the University Of Gent, Belgium [36]. The simulated Perovskite solar cell has layer configuration with transparent conductive oxide (TCO)/ blocking layer (TiO2)/ absorber/ and Si NWS as hole transport material. The considered materials for the mentioned layers are fluorine doped SnO2 (SnO2:F), pure TiO2 , CH3NH3PbI3 and spiro-OMeTAD (for reference cell), respectively. The thicknesses of layers are chosen based on experimental works on Perovskite solar cell [37]. To consider interface recombination, the interface layers INT1 and INT2 were defined from reference [38].
In this study, to obtain carrier diffusion lengths (Ln and Lp) of 1 μm that is similar to that of for experimental work [41], the  value of defect parameters for all layers are considered identical and defect density for absorber is assumed equal to Nt= 2.5 × 1013 cm3 [39]. 
Fig. 7 show I-V results for reference cell and cells with different Si NWs as HTMs. Also other important parameters are summarized at Table 2. From the results, reference cell shows efficiency of 20% VOC = 1.24 V, FF= 0.79, and JSC = 20.31 mA.cm-2, respectively.  When we used Si NWs with band gap of 1.1 ev which is equal to those in balk, efficiency, VOC, FF, and JSC dropped to 6.32 %, 0.91 V, 0.30, and 22.57 mA.cm-2, respectively.
By decreasing diameter of Si NWs, their band gap increased because of quantum confinement effect. As seen from the results, first, efficiency shows direct relation with increasing band gaps. When p-type Si NWs with band gap (60 nm in diameter) of 1.2 ev are used, efficiency of 21% are achieved. This cell has VOC, FF, JSC equal to 0.98 V, 0.80, and 23.11, respectively. Efficiency of 21.41 % is achieved when Si NWs with diameter of 5-7 nm was used. According to the P R Bandaru reports, its band gap is 1.4 ev [39].
As seen from the results, increasing band gap to 1.4 lead to decrease efficiency and JSC. By decreasing the diameter to 4 nm; consequently, band gap increase to 1.6 ev. In this case, efficiency, VOC, FF, JSC of 19.35 %, 1.07 V, 0.80, 21.20 mA.cm-2, respectively. Optimum band gap is 1.8 ev which is achieved by Si NWs with diameter of about 4 nm.  For this cell, related parameters are 21.1 %, 1.22 V, 0.89 and 22.14 mA.cm-2 for efficiency, VOC, FF, JSC, respectively. It seems that valence band of Si NWs determine efficiency. Fig. 8 shows related band diagram. As seen from the schematic, it is possible to tuning valance band and band gap of Si NWs by changing diameter of nanowires. Solar cells show better performance when valance band of Si NWs get closer to valance band of perovskite layer.
Electrical impedance spectroscopy (EIS) was adopted to evaluate charge transport and recombination in Si NWs based perovskite devices. Nyquist plot of these devices measured under dark with a 0.9 V bias is shown in Fig. 9. According to the equivalent circuit R1 indicate a series resistance, R2 interfacial resistance and R3 a recombination resistance, respectively.
Incident photon to converted electron (IPCE) results is presented in Fig. 10. Devices based on Si NWs with band gap of 2.2 ev show IPCE of less than 5 %. That why JSc of 2.5 mA.cm-2 was achieved for this cell. In other hand devices based on Si NWs with band gap of 1.2 ev shows highest incident photon to converted electron ratio. Almost 90% of incident photon with wavelength between 400- 700 nm converted to electron and hole pairs. For reference cell, this ratio is about 85% for same region
Our data shows that the series resistance almost is similar for all devices. Reference cell shows the lowest interfacial resistance while cell based on Si NWs with band gap of 1.8 ev shows highest recombination resistance. The larger interfacial resistance suggests that the Si NWs/ MAPbI3 interface is less efficient for charge transport. However, the larger recombination resistance of the Si NWs based suggests that the perovskite active layer is more efficient at suppress charge recombination, consequently enhanced device photovoltaic performance [39]. 

CONCLUSION
In conclusion, we report a new hole-transporting material (P-type Si NWs) to improve the stability of MAPbI3 film under ambient atmosphere and flow of water. MAPbI3 layer with was deposited on Si NWs with dip-coating method with grain size up to 3 µm and its degradation was monitored for 50 days. The champion p-i-n cell realizes very high stability even under water without any encapsulation. Our results shows that beside superior stability, it could exhibit very high efficiency exceed 21 %. The results provide an important facile approach to fabricate high-efficiency, stable and large area perovskite solar cell/module which accelerate the time to market. We propose three possible reasons for unusual stability of perovskite layer on Si NWS: first, hydrophobic properties of Si NWS. Si NWS has hydrophobic properties that repel water from its surface. Therefore water or moisture could not reach the surface of Si and therefore protect perovskite layer from water (contact angel for Si NWs is about 130 degree). Second, orientation of perovskite crystals changed over the time (this may happens to be stable against moisture) that why we named it, smart absorber layer. Third, as seen from the XRD results, perovskite crystals form very slowly on Si NWs (after 1 day).

ACKNOWLEDGMENT
This work was supported by kashan university, for supporting this work. The authors want to thank Dr. Ahmad Amiri from Tehran University for the helpful discussions.

CONFLICT OF INTEREST
The authors declare that there is no conflict of interests regarding the publication of this manuscript.