Versatile MoS2 Nanosheets in ITO-Free and Semi-transparent Polymer Power-generating Glass

Chemical exfoliated ultra-thin MoS2 nanosheets (NSs) with well 2D structure were demonstrated for interfacial layers and Ag nanowires composite transparent electrode in polymer solar cells (PSCs). The smooth and uniform n-type and p-type (after the plasma treatment) MoS2 NSs could improve fill factor of devices and light absorption in active layer. The optimized Ag nanowires–MoS2 NSs (AgNW-MoS2 NSs) transparent electrode presented a low sheet resistance of 9.8 Ω sq−1, and the corresponding transmittance also exhibited a high value of 93.1% at 550 nm. As a result, ITO-free PSCs based on AgNW-MoS2 NSs/n-MoS2 NSs cathode and p-MoS2 NSs/Ag anode achieved a highest PCE of 8.72%. Furthermore, a high efficiency (6.55%), large area and low cost semi-transparent power-generating glass was obtained, after reducing the thickness of top Ag electrode from 100 nm to 30 nm. To our best knowledge, it is the highest performance for semi-transparent PSCs devices reported up to now. The novel semi-transparent power-generating glass showed good performance and color purity for commercial applications in the near future.

A typical PSC is based on the bulk-heterojunction (BHJ) device configuration, which sandwiches a layer of polymer donor and fullerene acceptor blend between a transparent electrode (such as indium tin oxide (ITO)) and an opaque, reflective back electrode (such as aluminum (Al) or argentum (Ag)) 12 . There are also interfacial layers existing between active layer and electrodes, and the most common anode interfacial layer is poly (3, 4-ethylene dioxythiophene): poly (styrene sulfonic acid) (PEDOT: PSS) or MoO 3 for hole transportation and the ZnO or TiO x are the most widely used for cathode interfacial modification 7,[12][13][14] . A great effort is currently being exerted across a variety of research fields in the quest to achieve high-performance devices, for example, the synthesis of novel polymer donors, interfacial morphology control, optimized device structure and processing optimization. These devices have shown high power conversion efficiencies (PCE) approaching to 10% 4,[15][16][17] .
Unlike traditional bulk solar cells prepared from silicon or polysilicon, organic absorbing materials with confined absorption band can selectively be either transparent or semi-transparent in different regions of the solar spectrum 11,18 . Accordingly, semi-transparent polymer solar cells (STPSCs), an extension of PSCs which utilize transparent conductive materials as both electrodes, have recently gained much scientific attention and are considered to be the highest priority market for PSCs 9,19 . The STPSCs offer an extensive spectrum of applications such as power-generating windows for buildings, foldable curtains, mobile terminals and clothes, etc. To date, although many transparent polymer-based organic Scientific RepoRts | 5:12161 | DOi: 10.1038/srep12161 solar cells have been reported previously 18,[20][21][22][23][24] , the lack of appropriate device structure and suitable materials to use in interfacial layers and electrodes for PSCs has meant that these devices have exhibited either low efficiency (<5%, single junction ) or unsatisfactory average visible transmittance (AVT) and color purity for large-scale commercial applications 10,11 . Similar to graphene, other two-dimensional (2D) nanomaterials have been receiving great attention in recent years mainly due to their complementary electronic properties when compared with graphene [25][26][27][28] . Molybdenum disulfide (MoS 2 ) is a 2D material consisting of hexagonal sheets of molybdenum (Mo) sandwiched between two hexagonal sheets of sulfur (S). MoS 2 exhibits specific anisotropic mechanical and electrical properties due to the weak bonding between layers 29,30 . A mobility value of over 200 cm 2 V −1 s −1 has been reported in a field-effect transistor architecture 31,32 . It has also been demonstrated that MoS 2 could act as Schottky-barrier active layer or hole transport layer (HTL) in photovoltaic [33][34][35][36][37] . However, MoS 2 has not been taken full advantages of its 2D plane properties, even though there are few traps on the layer surface due to its lack of dangling bonds.
We report here on materials and processes for reliable and cost-efficient processing of ITO-free semi-transparent polymer solar cells by introducing ultra-thin versatile MoS 2 nanosheets as interfacial layers and semi-transparent electrodes with employing a silver nanowires-n-type MoS 2 nanosheets (AgNW-MoS 2 NSs) cathode and thin Ag metal anode. This is the first time to report a novel AgNW-MoS 2 NSs/n-MoS 2 NSs composite cathode can form excellent 2D morphology for interfacial contact and optical distribution in devices, and p-MoS 2 NSs/ thin Ag metal is found to achieve a suitable anode with high conductivity and transparency at the same time. The ultra-thin versatile MoS 2 can effectively avoid aggregation of AgNW electrode and is beneficial for interfacial modification of devices, proposing the MoS 2 can replace the traditional electrode and interface layer in organic optoelectronic materials. Incorporation of the AgNW-MoS 2 NSs/n-MoS 2 NSs cathode and p-MoS 2 NSs/ thin Ag as the hybrid anode not only enhances fill factor of devices but also improves the light absorption in the BHJ layer and the transmittance of the devices simultaneously as shown in Fig. 1. After optimization of the devices, combination of material selection and device engineering facilitates the fabrication of ITO-free PSCs and STPSCs with good average performance of 8.00% (best value of 8.72%) and 6.02% (best value of 6.55%), respectively, based on poly [4,8- [3,4-b]thiophen-4,6-diyl] (PBDTTT-C-T) 38 . (6,6)-phenyl-C 70 butyric acid methyl ester (PC 70 BM) (PBDTTT-C-T:PC 70 BM) system. Furthermore, these good chrominance large-scale STPSCs module based on the novel device structure can light a LED chip under natural light illuminated, which shows great application possibility in power-generating windows for building integrated photovoltaic (BIPV).

Methods
Sample Preparation. Chemically exfoliated Molybdenum sulfide (MoS 2 ) nanosheets (NSs) were synthesized through Li intercalation. 0.8 g raw MoS 2 powders (Sigma-Aldrich) were immersed in 24 mL n-butyllithium (n-BuLi, 2.5 M hexane solution) solution. The mixture was stirred at room temperature for 48 h under an Ar atmosphere and then allowed to settle for hours. The mixture was centrifuged several times to remove LiOH and unexfoliated material. The obtained MoS 2 NSs were n-type materials, for the p-type MoS 2 NSs were then prepared after 30 min oxygen plasma treatment (PDC-32G-2 PLASMA CLEANERM, high level, 18W). The sample was diluted with isopropyl alcohol to ~0.5 mg mL −1 for device fabrication.
For the post-treatments, the AgNWs-MoS 2 ink was manufactured by blending the Ag NWs dispersion (diam. × L 60 nm × 10 μ m, 0.5% isopropyl alcohol suspension Sigma-Aldrich) and n-type MoS 2 isopropyl alcohol solution (1:1,V:V). Preparation of the Ag NWs and AgNWs-MoS 2 Films: The Ag NW films were prepared using the spin-coating process and were formed on precleaned glass substrates that were attached to a supporting glass substrate. The as-received dispersion containing Ag NWs and AgNWs-MoS 2 were spin coated for 40 s at speeds ranging from 500-4000 rpm. The dispersion was sonicated for 30 min and shaken well before being the spin-coating process. The formed electrode films were annealed at ≈ 120 o C for 10 min in a glove box filled with nitrogen (the detail process was shown in support information in Supplementary Fig. S1).
Device Fabrication. The polymer solar cells were fabricated on Glass/ITO, Glass/AgNW or Glass/Ag NW-MoS 2 electrodes. For the electron transport layer, the sol-gel derived ZnO film using the reported method 39 or 40 nm thickness of n-type MoS 2 was placed on the top of electrode. The thickness of ZnO film is approximately 30 nm, determined by a profilometer (Alpha-Step-IQ). Subsequently, the modified samples were transferred to the nitrogen-filled glove-box. The D-A copolymer-containing photoactive layer was prepared by spin coating (900 rpm) the dichlorobenzene solution of D-A copolymer (PBDTTT-C-T) and PC 70 BM (1:1.5 w/w, polymer concentration of 10 mg mL −1 ) with 3% volume ratio of DIO additive on the modified electrode. The thickness of the photoactive layer was about 100 nm. Then, the MoO 3 precursor or p-type MoS 2 solution with optimal thickness was spin-cast on top of the polymer:fullerene composite layer. Finally, the device was pumped down in vacuum (< 10 −7 torr; 1 torr~133 Pa), and a~100 or 30 nm thick Ag electrode was deposited on top. The deposited Ag electrode area defined the active area of the devices as 0.06 or 0.65 cm 2 (all the areas were tested with an aperture).
Power-generating glass Fabrication. The power-generating glass was fabricated from eight 0.65 cm 2 (effective area) Device E. All the devices were weld together, and then encapsulating by UV curing glue. The LED chip was accessed in the circuit.
Characterization. X-ray diffraction (XRD) patterns of the MoS 2 nanoparticles were carried out on a Bruker D8 Focus X-ray diffractometer operating at 30 kV and 20 mA with a copper target (λ = 1.54 Å) and at a scanning rate of 1 o/min. Sheet resistances of electrodes were measured by using a four point probe setup with a source measurement unit (Keithley 2400). Current-voltage (J-V) characteristics were characterized using Keithley 2400. The currents were measured in the dark and under 100 mW·cm −2 simulated AM 1.5 G irradiation (Abet Solar Simulator Sun2000). All the measurements were performed under ambient atmosphere at room temperature. The scan range is form 0 V to 1V, and 6.7 mV for each step. All the J-V curves are based on 150 points. The incident photo-to-electron conversion efficiency spectrum (IPCE) were detected under monochromatic illumination (Oriel Cornerstone 260 1/4 m monochromator equipped with Oriel 70613NS QTH lamp), and the calibration of the incident light was performed with a monocrystalline silicon diode. Transmittance spectra were analyzed by UV-vis spectroscopy (Perkin Elmer Lambda 750). The morphologies of films were investigated by atomic force microscopy (AFM) using a Digital Instrumental Nanoscope 31 operated in the tapping mode. The thicknesses of all the layers were measured by surface profilometer (Alpha-Step-IQ). XPS studies were performed on a Thermo-VG Scientific ESCALAB 250 photoelectron spectrometer using a monochromated AlKa (1,486.6 eV) X-ray source. All recorded peaks were corrected for electrostatic effects by setting the C− C component of the C 1s peak to 284.8 eV. The base pressure in the XPS analysis chamber was 2 × 10 −9 mbar. For the UPS measurements, He I (21.22 eV) radiation line from a discharge lamp was used, with an experimental resolution of 0.15 eV. All the UPS measurements of the onset of photoemission for determining the work function were done using standard procedures with a − 5 V bias applied to the sample. Raman spectroscopy was performed using an InVia Raman Microscope system (Renishaw, Inc.), with an Ar+ ion laser operating at 613 nm and 1.2 mW. All the simulation were calculated via Matlab program, the Matlab files were provided from McGehee group, Center for Advanced Molecular Photovoltaics, Stanford University. For the evaluation of the color rendering indices (CRIs), the experimental transmission of each device is folded with the AM1.5 spectrum to obtain the perceived transmission under solar illumination. The resulting data is coupled with the CIE 1931 2 o standard observer color matching functions to obtain the corresponding xyY points.

Results
Properties of ultrathin MoS 2 NSs. In order to obtain ultrathin MoS 2 NSs, the raw MoS 2 powders were exfoliated by chemical lithium intercalation-exfoliation method 28,40 . Generally, the synthetic procedure involves two steps: n-BuLi can intercalate bulk MoS 2 with lithium atoms and form LixMoS 2 (0 < x < 1) in the gram scale or more, at room temperature. And then LixMoS 2 can react with H 2 O rapidly and release large amounts of hydrogen gas, which can push the MoS 2 layers and cause them to separate from each other and form homogeneous ultrathin MoS 2 NSs. The powder X-ray diffraction (XRD) patterns of raw and exfoliated MoS 2 nanosheets ( Supplementary Fig. S2, Supporting Information), after chemical exfoliation, show that only the peaks of (002) and (103) plane remain which confirms that the MoS2 nanosheets were successfully striped. And the disappearance of other peaks could proves ultrathin MoS 2 nanosheets tightly lie on the substrate with preferred orientation 35 Table 1 and Fig. 3a and Supplementary Fig. S6. The optical and electrical parameters of pristine AgNW electrode are basically identical with the previous reports. It is well known that a competitively transparent electrode must has a transparency of at least 90% and a sheet resistance of less than 10 Ω /sq 43 , therefore only AgNW ink is hard to satisfy the two requirements at the same time 44,45 . Delightfully, the modified AgNW-MoS 2 loading shows a relative low sheet resistance of 9.8 Ω sq −1 , and the corresponding transmittance of the electrode also exhibits a high value of 93.1% at 550 nm. It is obvious to found that the transmittance of the electrodes via different speed after the addition of MoS 2 NSs has a significant increase, mainly attributed to its high refractive index of MoS 2 NSs and a well 2D plane structure. Moreover, the environmental stability of AgNW and AgNW-MoS 2 electrodes (500 rpm spin-coating speed) were tested under 50 o C and 70 relative humidity (RH) conditions for 7 days, as shown in Supplementary Fig. S7. AgNW electrode is easy to oxidation under the atmospheric environment 46,47 , nevertheless, the incorporation of MoS 2 NSs in AgNW electrodes show eminent oxidation resistance and moisture absorption stability.
The pristine MoS 2 NSs are n type semi-conductor materials 34,48 , several work has been reported that MoS 2 could be changed as a p type semi-conductor materials with a relative high work function after UV-ozone plasma treatment 34,37,48 . Thus, the properties of MoS 2 NSs before or after the UV-ozone plasma treatment were also investigated. Supplementary Fig. S8 is the X-ray photoelectron spectroscopy (XPS) profiles of n-MoS 2 NSs (w/o plasma treatment) and p-MoS 2 NSs (with plasma treatment). The Mo 3d . Therefore, it is proved that n-MoS 2 NSs can be successfully oxidized to p-type materials after plasma treatment 35 .
The work function (WF) of the n and p type MoS 2 NSs was determined via UPS ( Supplementary Fig.  S9) and Kelvin probe ( Supplementary Fig. S10). The WF of the n-MoS 2 NSs (pristine MoS 2 NSs) is found to be 4.3 eV, while that of the AgNW-MoS 2 electrode is 4.1 eV, which is even lower than Ag electrode (4.3 eV), meaning more efficient electron collection for cathode 12 . As expected, after plasma treatment, the oxygen doping at the surface of the MoS 2 NSs film and making the WF of p-MoS 2 NSs upshifts to 5.0 eV, which is nearly to the WF of MoO 3 (5.3 eV). The substantial increased work function is close to the HOMO level of PBDTTT-C-T, as illustrated in Fig. 3b, which can avoid the hole from capturing and gathering at the interface to form a recombination center, leading to more effective hole selection and transportation for PSCs 12 .
Photovoltaic performance and characterization. Besides suitable energy alignment in the devices, excellent interfacial morphology is also critical for high performance PSCs 13 . The formation of the AgNW and AgNW-MoS 2 films can also be verified by optical microscope (OM) and AFM images displayed in Supplementary Fig. S11 and Fig. 4a,b. Enormous aggregation of AgNW can be observed from large-scale OM and micro-scale AFM observation, while the AgNW-MoS 2 film develops more smooth morphology with no apparent aggregation in Supplementary Fig. S11. Intriguingly, the AFM image of AgNW-MoS 2 film demonstrates highly oriented Ag nanowires arrange uniformly in MoS 2 NSs background, with a more smooth surface than AgNW one (RMS:1.68 vs 21.62, ). These oriented Ag nanowires can provide a convenient pathway for electrons transportation and collection in AgNW-MoS 2 electrodes, beneficial to the high conductive electrode. In addition, the interpenetrating network structure of AgNW and MoS 2 NSs also render a great contribution to the homogeneous surface of AgNW-MoS 2 film and high transmittance. The Ag composite electrodes could also have a great influence on active layer, as shown in Supplementary Fig. S12. Supplementary Fig. S12a shows the morphology of BHJ active layer which based on AgNW electrode with a rough surface (RMS: 6.46). However, the AgNW-MoS 2 film can provide a 2D substrate for deposition of BHJ active layer with a reduced RMS of 1.21. The different BHJ active layers and p-MoS 2 NSs have a great effect on the morphology of top Ag electrode as well. The AFM height and three-dimensional images of 30 nm Ag film deposited on bare BHJ layer (Fig. 4c,e), display a horrifically rough surface with a RMS of 20.07, leading to the poor device performance (discuss later). In contrast, the Ag film covered on p-MoS 2 layer/BHJ exhibits an uniform and smooth surface morphology in Fig. 4d,f, which means p-MoS 2 NSs can act as a soft mattress and provide an ingenious interfacial contact between BHJ layer and Ag electrode.
To  interfacial transportation to enhance the charge extraction and reduce the charge combination, leading to the improvement in the J sc . When the ITO cathode was replaced to AgNW, all of the device parameters, e.g. PCE, J sc , V oc and FF are substantially reduced in Device C (Glass/AgNW/n-MoS 2 /BHJ/p-MoS 2 / Ag(100 nm)), resulting from the poor transmittance of cathode and reduced light absorption in active layer. Intriguingly, after introducing AgNW-MoS 2 electrode, the PCE of ITO-free device is further improved from 6.39% for Device C to 8.00% for Device D (Glass/AgNW-MoS2/n-MoS2/BHJ/p-MoS2/ Ag(100 nm)), which is even comparable to the control Device B based on ITO cathode. The improved efficiency is mainly related to the increased J sc (15.66 mA cm −2 ) and FF (0.66), revealing the function of AgNW-MoS 2 electrode and in good agreement with AFM measurement (Fig. 4 and Supplementary  Fig. S12).
To fabricate a semi-transparent polymer solar cells (STPSCs), the top Ag electrode directly decreased Ag thickness from 100 nm to 30 nm 11 , since the control device with 100 nm thick Ag anode was almost opaque (average visible-light transmittance: < 0.05). When the thickness of top Ag layer decreases to 30 nm for a thin electrode, the total transmittance of the devices shows an average value of 0.21, as shown in Supplementary Fig. S13. Gratifyingly, the STPSCs based on the structure (Device E: Glass/ AgNW-MoS 2 /n-MoS 2 /BHJ/p-MoS 2 /Ag(30 nm)) emerges an average PCE of 6.02%, with a J sc of 12.66 m −2 , an V oc of 0.75V and a FF of 0.64, when the light was illuminated from bottom side, the PCE is the highest values reported so far for ITO-free semi-transparent PSCs devices. Moreover, the PCE of the Device F with the light injected from the top Ag side also present a PCE of 2.77%. The effectiveness of Device E structure inspires us to apply it for large-scale STPSCs. The J-V curves of the STPSCs which increase the effective area from 0.06 cm −2 to 0.65 cm −2 were shown in Fig. 5b and Table 2. the ITO-free inverted device with a structure of Glass/AgNW-MoS 2 /n-MoS 2 /BHJ (0.65 cm −2 )/p-MoS 2 /Ag (30 nm) ) (Device G) still shows a PCE as high as 2.71%, with a V oc of 0.54 V, a J sc of 9.38 mA·cm −2 , and an FF of 0.53, and parameters of the device tested from top Ag side also reach a high level (1.20, Device H). The IPCE spectra of different devices based on PBDTTT-C-T:PC 70 BM are shown in Fig. 5c,d, and the values of J sc via the IPCE measurement well match with the ones from I-V curves. These results demonstrate the universality of this novel and large-scale STPSCs.

Discussion
To find out the reason for higher efficiency of PSCs caused by introducing the MoS 2 NSs interfacial layer, the simulation of the spatial distribution of the squared optical electric field (normalized to the incoming plane wave) for the devices with ZnO/MoO 3 and n-MoS 2 and p-MoS 2 were shown in Fig. 6a. The increased J sc of these devices based on MoS 2 interfacial layers can be attributed to high refractive index of n-MoS 2 NSs (3.25 from 300 nm to 800 nm) leading to the higher light intensity within the active layer, as shown in Fig. 6a. From the figure we also can see that there is a dramatic improvement of the optical electric field intensity of 500 nm wavelength in active layer of the devices with n-MoS 2 /p-MoS 2 with respect to the devices with ZnO/MoO 3 . Furthermore, the simulation of the spatial distribution of the squared optical electric field for the Device C, D and E were presented in Fig. 6b, where the figures of 0.36, 0.37 and 0.28 in brackets mean the relative optical electric field intensity of active layer for different devices. It is obviously found that there is no big difference between Device C and D (0.36 vs 0.37, the relative area ratio). The thickness of the top Ag electrode sharply decreases from 100 nm to 30 nm, while  the relative optical electric field intensity of active layer slightly reduces to 0.28 for Device E. In addition, to gain deeper insight into the performance of the large-scale STPSCs were illuminated under different light intensity, we studied a power law dependence of J sc upon illumination intensity is generally observed in PSCs and can be demonstrated as:  where I is the light intensity and α is the exponential factor 50,51 . In Supplementary Fig. S14, the data are plotted on a log− log scale and fitted to a power law using eq 1: α = 0.955, higher than the reference device 52 , indicating that bimolecular recombination could be less in the cell with the MoS 2 NSs transport layer. It is also found that the STPSCs based on Device E structure behave an excellent performance under weaker illumination, as shown in Supplementary Fig. S14.
To evaluate the transparency perception of the above STPSCs in human eyes, the tristimulus value (X, Y, Z) and the color coordinates (x, y) were calculated from the transmission spectra while the incident light source was replaced with the AM 1.5 spectrum. The color coordinates of the studied STPSCs with different thickness of Ag on CIE 1931 chromaticity diagram are illustrated in Fig. 7a  (0.333, 0.347), and (0.313, 0.329), respectively. All these films show very nice transparency color perception that is close to AM 1.5 point. Fig. 7c,d show two digital photographs taken through and not taken through the STPSCs based on Device E structure. There is no big aberration between the two pictures, which means it could provide an opportunity for these STPSCs applying in power-generating windows. In order to further to prove the possibility, a power-generating glass was fabricated which is made up of eight 0.65 cm 2 Devices E, as shown in Fig. 8. Fig. 8a delivers the power-generating glass illuminated under 1.5 AM solar condition and the LED chip basically can achieve the rated power. Miraculously, the solar cell glass presents an excellent performance out of doors condition (Test time: October 25 th , 2014 PM 13:00 and Environmental conditions: Nanchang City, Jiangxi Province, P.R China, 25 o C Sunshine and 43% HR ). The photocurrent of power-generating glass is enough to light the LED chip regardless of illumination direction from the front or back side. From these applied tests, it can be sure to say, this kind of power-generating glass provide a great opportunity for building integrated PV (BIPV).

Conclusions
In summary, we have demonstrated a versatile MoS 2 nanosheets with well 2D structure that were introduced into Ag nanowires transparent electrode, anode and cathode interfacial layers in PSCs. High performance ITO-free PSCs based on Glass/AgNW-MoS 2 /n-MoS 2 /BHJ/p-MoS 2 /Ag(100 nm) achieve a high PCE of 8.00%.
Furthermore, a novel STPSC is incorporating the MoS 2 NSs in interfacial layers and AgNW transparent electrode with admirable interfacial contact and superior light-harvesting abilities. The STPSCs exhibited good PCE and AVT with neutral color perception close to that of standard sunlight, which is the highest performance for ITO-free semi-transparent PSCs devices reported up to now. The delightful results demonstrate STPSCs do have great values in power-generating windows of building integrated PV in near future.

Figure 8. Power-generating glass images.
Power-generating glass was fabricated via combine eight 0.65 cm 2 deivces E, the devices were encapsulated by UV curing glue. To test the power of this semi-transparent solar cell glass, a 5 mW LED chip was put in circuit. (a) the power-generating glass was illuminated under 1.5 AM solar condition (b) the power-generating glass was tested from back side out of doors (c) the powergenerating glass was tested from front side out of doors.