Low-temperature , solution-processed , layered V 2 O 5 hydrate as the hole-transport layer for stable organic solar cells †

Layered V2O5 hydrate has been applied as the hole transport layer (HTL) in organic solar cells (OSCs). V2O5 is obtained from a sodiummetavanadate solution in water under ambient conditions, resulting in a final thin film of formula V2O5$0.5H2O. The 0.5 water molecules are not removed from the V2O5 layered structure unless the sample is heated above 250 C, which makes the thin film highly stable under real working conditions. The HTL was used in OSCs in the normal and the inverted configurations, applying metallic Ag as the back-metal electrode in both cases. Fabrication of both OSC configurations completely by solution-processing printing methods in air is possible, since the Al electrode needed for the normalconfiguration OSC is not required. The work function (WF) and band gap energy (BG) of the V2O5 thin films were assessed by XPS, UPS and optical analyses. Different WF values were observed for V2O5 prepared from a fresh V2O5–isopropanol (IPA) solution (5.15 eV) and that prepared from a 24 h-old solution (5.5 eV). This difference is due to the gradual reduction of vanadium (from V to V) in IPA. The OSCs made with the V2O5 thin film obtained from the 24 h-old V2O5–IPA solution required photoactivation, whereas those made with the freshly obtained V2O5 did not. Outdoor stability analyses of sealed OSCs containing a V2O5 HTL in either configuration revealed high stability for both devices: the photovoltaic response at T80 was retained for more than 1000 h.


Introduction
The predicted maximum power conversion efficiency (PCE) of organic solar cells (OSCs) has been empirically estimated to be 10 to 12%, but theoretical calculations suggest that values of 20 to 24% could be achieved, which are comparable to those of crystalline Si solar cells. 1 However, OSCs must also be costcompetitive and show long lifetimes.3][4][5][6][7][8] Ideally, a single printing method would be used to continuously and rapidly process all layers; unfortunately, no such process exists yet and therefore, multiple techniques must be employed.Additionally, the envisaged mass-production of OSCs indicates that toxic organic solvents will have to be replaced with non-toxic, alcohol or water-based solutions and inks.][11] Transition metal oxides (TMOs) have been employed in organic solar cells, especially TiO 2 and ZnO.Their most attractive feature is the possibility to be deposited by low temperature solution processing methods.Among them are also V 2 O 5 , [12][13][14][15][16][17][18][19][20][21] NiO, [22][23][24][25][26][27][28][29][30] MoO 3 , [31][32][33][34][35] WO 3 [36][37][38][39] or Sb 2 O 3 . 40These TMOs exhibit a wide range of energy level alignments, [41][42][43][44] good transparency as thin lms, are easy to manipulate, and confer low-resistance ohmic contacts to the OSC. 454][55][56] One interesting TMO is V 2 O 5 , which has been reported to be a good candidate for the HTL in OSCs.To date, V 2 O 5 HTLs have been synthesised chiey by multistep techniques.Examples include the suspension of V 2 O 5 nanoparticulates obtained from the hydrolysis of vanadium(III) acetyl acetate 19 or the fabrication of a bronze V 2 O 5 HTL from a suspension of the metal oxide obtained aer the reaction between the metal powder and H 2 O 2 . 21Among the most widespread fabrication methods is the application of sol-gels made from vanadium(V) oxytriisopropoxide (ViPr), 5,14,15,57,58 which is a compound known for its high toxicity, reactivity and cost.
Herein we present the synthesis, optimisation and application of water-based, solution processable V 2 O 5 as the HTL in OSCs.The HTL was fabricated at low temperature in air without the need for any high temperature post-deposition treatments or multistep reactions.The water-based layered V 2 O 5 hydrate is highly compatible with the fabrication of OSCs by large-area, low-cost, fast processing and high-throughput printing, 5,15 and also enables the preparation of ZnO/V 2 O 5 recombination layers required for TmOSCs. 59We demonstrate here that the application of the low-temperature water-based V 2 O 5 solution can be tuned in order to fabricate OSCs in either the inverted or the normal congurations, on either glass or exible substrates.Moreover, we have fabricated OSCs with both congurations applying only Ag as the back metal electrode.Thus, our OSCs can be made completely by solution-processing methods, as they do not require an Al electrode for the normal-conguration OSCs 10 and the Ag metal electrode can be deposited by established solution-processing printing techniques. 11A careful optimisation of the V 2 O 5 hydrate solution permitted to obviate the requirement for photo-activation of the solar cell in air. 19inally, the OSC devices also show good outdoor stability maintaining T 80 for more than 1000 h.

Synthesis of TMOs
The TiO 2 solution was fabricated and deposited as previously reported, 60 and the ZnO nanoparticles were synthesised following the Pacholski method 61 and deposited by spin coating at 1000 rpm.The V 2 O 5 hydrate solution was obtained using an adapted version of Livage's method. 62Briey, 4.5 g of sodium metavanadate (NaVO 3 ) are dissolved in 100 mL of deionised water by heating the mixture at 90 C with stirring.The resulting transparent solution is passed through a cation-exchange resin (DOWEX 2x-100, Aldrich) to obtain metavanadic acid (HVO 3 ) as a yellow solution, which is stored in a closed glass ask under an ambient atmosphere.As the solution ages, it becomes yellowish-orange, indicating the presence of condensed species such as decavanadates ([V 10 O 28 ] 6À ).Finally, the orange solution changes to dark red, which is characteristic of the V 2 O 5 hydrate (gel).The condensation and aging of the solution occurs over $20 days, aer which, the viscous hydrate solution stabilises and is ready to use.A bright red vanadium pentoxide V 2 O 5 $nH 2 O gel is obtained.For the inverted OSC a 1 : 1 mixture of V 2 O 5 hydrate (9 mg mL À1 ) in isopropanol (IPA) was spincoated at 1000 rpm, and then annealed at 120 C for 5 min.For the normal conguration OSCs the V 2 O 5 solution was spin coated at 1000 rpm directly from the solution without the aid of the IPA.

Solar cell fabrication
The response of the solar cells was independent of the used substrate (ITO or FTO).All the exible substrates were made on a PET/ITO transparent lm.Substrates of FTO on glass were supplied by SOLEMS (resistance: 70 to 100 ohms).PET/ITO substrates were brought from Aldrich (resistance: 35 ohms).The substrates were cleaned with a water-soap solution, rinsed with deionised water, ultrasonicated in ethanol (99%), dried under N 2 ux and nally, ozone treated in a UV-surface decontamination system (Novascan, PSD-UV) connected to an O 2 supply.The OSCs were fabricated in either the normal conguration (FTO/V 2 O 5 /P3HT:PCBM/TiO 2 /Ag) or the inverted conguration (FTO/TiO 2 /P3HT:PCBM/V 2 O 5 /Ag).TiO 2 was used on glass/FTO substrates and was replaced by ZnO in order to fabricate exible OSCs on ITO/PET (Aldrich).The bulk heterojunction blend comprised a 1 : 0.8 mixture of regioregular P3HT (Merck, 98%) and PCBM (Solenne, 99.5%) that was dissolved in dichlorobenzene (Sigma-Aldrich).The blend was spin coated at 1000 rpm, affording a 200 to 250 nm-thick P3HT:PCBM lm.The devices were subsequently annealed at 120 C for 5 min in air.A 100 nm-thick Ag back metal electrode was deposited by thermal evaporation in an evaporation system (Auto 306, Broc Edwards) with a base pressure of 10 À7 torr at a deposition rate of 1 Å s À1 .The nal active area was 0.2 cm 2 .The solar cells were sealed by applying a two-component adhesive (ThreeBond, 30Y-727 and 31x-167-2), which was mixed in a 1 : 1 ratio and cured under UV light for 10 min.

Characterisation
The thicknesses of the polymer layers were measured using a Nanopics-2100 from Nanopics prolometer and by SEM (Quanta FEI 200 FEG-ESEM).Optical measurements were performed using a UV-vis spectrophotometer (UV 1800, Shimadzu).Grazing incidence X-ray analyses were done using a Rigaku unit and measured between 5 and 80 .TGA analyses were done using an STA 449 F1 Jupier (Netzsch).Contact angles were analysed using a DSA 100 (KRUSS).XPS was done with the Al ka (1486.6 eV).All spectra were adjusted according to the value of the C 1s peak at 284.4 AE 0.1 eV.The UPS were obtained using a He lamp (He l 21.2 eV) at an experimental resolution of 0.15 eV.The samples were biased at À5 V.

Paper
Energy

Photovoltaic characterisation
The solar simulation was performed on a KHS1200 (Steuernagel Solarkonstant) equipped with an AM1.5 lter for all characterisations (100 mW cm À2 , AM1.5G, 72 C).The equipment was calibrated according to the ASTM G173.IV-curves were measured using a Keithley 2601 multimeter.Light intensity was 100 mW calibrated with a Zipp & Konen CM-4 pyranometer, which was used constantly during measurements to set light intensity, and a calibrated S1227-1010BQ photodiode from Hamamatsu was also applied for calibration before each measurement.IPCE analyses were done with a QE/IPCE measurement System from Oriel (from 300 to 700 nm; at 10 nm intervals).The results were not corrected for any intensity losses due to light absorption or reection by the glass support.

Outdoor stability analyses
The outdoor stability analyses were done following the ISOS-1 procedures 63 at the Laboratory of Nanostructured Materials for Photovoltaic Energy of the Catalan Institute of Nanoscience and Nanotechnology (ICN2-CSIC), located in Barcelona, Spain (41.30N 2.09 W), using a solar tracking positioning system.
The system comprises a large dual axis-controlled platform with fully automated motors, which enables turning of the tracker hour angle up to 100 (which translates to nearly 7 hours of perpendicular solar tracking) and turning of the tracker elevation angle from 15 to 90 (which enables full tracking of solar elevation).We developed in-house soware to control the photovoltaic response of sixteen solar cells at the same time and to continuously monitor light irradiation, temperature and relative humidity over time.IV-curves were measured using a 2602A dual-channel SMU multimeter and a 3700 series switch/ multimeter (both from Keithley).PCE values were calculated using the maximum daily irradiance level.The light irradiation was measured with a Zipp&Konen CM-4 pyranometer.The temperature and relative humidity were monitored with a combined sensor (Theodor Friedrichs).
3 Results and discussion The V 2 O 5 lm used in this work was prepared by spin coating and the substrate was then treated at 120 C for several minutes before being applied to the OSCs.The nal formula of the thin lm is V 2 O 5 $0.5H 2 O (the number of water molecules was calculated from the TGA analyses performed in air).Once the lms have been prepared, the water molecules can only be removed if the lm is heated above 250 C.However, under real working conditions the OSCs will never reach those temperatures and therefore, the water in the V 2 O 5 interlayer can be considered stable.
Fig. 1 shows the photovoltaic response of inverted organic solar cells with the glass/FTO/TiO 2 /PCM:P3HT/V 2 O 5 /Ag conguration, depending on the V 2 O 5 concentration.For the fabrication of inverted devices, the deposition of V 2 O 5 on top of the active P3HT:PCBM layer requires mixing of V 2 O 5 hydrate with isopropanol (IPA) to improve adherence.The optimum ratio of V 2 O 5 to IPA was found to be 1 : 1 (as determined by contact angle measurements; see ESI Fig. S1 †).The solution obtained was then spin-coated at 1000 rpm in an ambient atmosphere, and nally, heated at 120 C for 5 min.There is a clear improvement on the photovoltaic response of the device (ca. in FF, J sc and PCE) when the concentration of the oxide is increased from 2 mg mL À1 to 9 mg mL À1 , as observed in Fig. 1.The PCE and FF increase and ultimately stabilise at 3% and 50%, respectively.As the FF increased, the J sc also stabilised at a V 2 O 5 concentration of 9 mg mL À1 .Increasing the photovoltaic response by increasing the concentration of V 2 O 5 above these values was not achievable due its limited solubility.Thus, the optimal value chosen for fabrication of the inverted OSCs was 9 mg mL À1 .An increase in the thin lm layer thickness was also observed when raising the concentration of the oxide from 2 mg mL À1 to 9 mg mL À1 , with values ranging from 60 nm to 125 nm (as measured by SEM and prolometry, Fig. 1e).The possibility to fabricate thick lm layers without compromising the photovoltaic performance of the device (i.e. by increasing series resistance) has also been observed by Brabec et al. 19 In our case, we attributed this response to the mixed ionic-electronic conductivity that characterises the V 2 O 5 $0.5H 2 O thin lm.Finally, Fig. 1f shows the atomic force microscopy (AFM) image, at a 2 mm Â 2 mm scan size, of the thin lm V 2 O 5 made with a solution concentration of 9 mg mL À1 .A nanostructured surface with high surface roughness is observed for the thin lm.
To compare the performance of our V 2 O 5 HTLs with that of the most widely used HTL, PEDOT:PSS, we fabricated and assessed OSCs of both congurations.The devices were prepared on glass/FTO substrates (the fabrication of exible OSCs applying the V 2 O 5 HTL is also possible, see ESI Fig. S2 †).Fig. 2 shows the IV curves and the IPCE spectra obtained for the devices.In all cases the ETL was ZnO.Table 1 shows the photovoltaic parameters obtained for the different OSCs; for comparison purposes, we have also included solar cells containing a TiO 2 ETL.The reported values are mean values from six samples.The best photovoltaic performance (PCE: 3%) was generally observed for the OSC fabricated on glass/FTO substrates, with TiO 2 as the ETL and V 2 O 5 hydrate as the HTL.In this case, the OSCs employing the V 2 O 5 hydrate resulted in better performance when compared to PEDOT:PSS.The OSCs with ZnO and V 2 O 5 showed a very similar response with photovoltaic PCEs of ca.2.5 to 2.6%.Our results indicate that a similar response can be achieved for OSCs with layered V 2 O 5 hydrate when compared to the PEDOT:PPS HTL.

S-shape curve and photo-annealing of the inverted OSC in air
Fig. 3 shows the IV-curves and IPCE spectra obtained for our OSCs fabricated with fresh (1) and 24 hour-old (2) V 2 O 5 -IPA solutions.As observed, the OSC made with the fresh solution exhibited maximum photovoltaic performance directly aer fabrication (Fig. 3a-(1)).In contrast, the OSC made with the 24 h-old solution (Fig. 3a-(2)) gradually improved with IV-cycles, ultimately reaching maximum performance (Fig. 3a-(3)).The corresponding IPCE analyses (see Fig. 3b) are in close agreement with the PCE values obtained for the OSCs.Besides the difference in IPCE intensity values, the most signicant variations between the IPCE applying freshly prepared (1) and 24 h-old (2) V 2 O 5 -IPA solutions are observed in the wavelength region below 450 nm, corresponding to the adsorption of the semiconductor oxides (TiO 2 , ZnO, V 2 O 5 , etc.).
Since the only difference in the fabrication of the two OSCs was the V 2 O 5 HTL, we attributed the need for photo-annealing to the V 2 O 5 thin lm properties that are probably affected by the interaction between V 2 O 5 with IPA. 62,66,67Layered vanadium(V) oxides in their hydrated state tend to accommodate foreign molecules in their interlayer region, 64,[66][67][68] including organic compounds such as alcohols. 69Alcohols intercalate, via their -OH group, at the polar site of V 2 O 5 .In this partially reversible reaction the H 2 O molecules of V 2 O 5 hydrate are exchanged with alcohol molecules, leading to the reduction of V 2 O 5 from V +5 to V +4 .V 2 O 5 reduces relatively quickly when in solution with organic molecules, as indicated by a gradual change in the colour of the solution from red (indicative of V 5+ ) to green (indicative of the reduction of V 5+ to V 4+ ).Thus, the thin lm obtained from the 24 h-old V 2 O 5 -IPA solution could be partially reduced and therefore photo-annealing is required in order to eliminate the undesirable shunts and inection points (S-shape IV curve) and to achieve maximum power conversion efficiency. 4,70,71,807][78] Fig. 4a shows the XPS spectra of the V 2 O 5 $0.5H 2 O thin lms fabricated from the fresh (red) and 24 h-old (green) solutions.The binding energy (BE) values of the main peaks and their assignment are detailed in Table 2.
XPS analyses revealed only slight differences in the intensity of the spectra between the two lms (see ESI Fig. S3 †).Despite these small differences, the two thin lms gave very similar XPS results: the main peaks of V2p 3/2 and V2p 1/2 were almost identical.The characteristic peaks of V 2 O 5 are observed at 517 eV and 524 eV (corresponding to V 5+ ), and at 529.9 eV (the O 1s from the O 2À ions).The XPS plot was subject to a Lorentzian-Gaussian tting: the region of the V 5+ peak at 517 eV reveals a shoulder at ca. 516 eV.This peak is attributed to the presence of V 4+ , which is commonly observed in the hydrated form of V 2 O 5 72 as well as in reduced lms. 73But, it is not present in a crystalline V 2 O 5 lm that has been subjected to thermal evaporation or annealed at high temperatures, as these procedures eliminate all water. 72This shoulder at 516 eV has also been observed by Ziberberg et al., in the XPS analyses of a V 2 O 5 thin lm (10 nm) obtained from vanadium(V)-oxytriisopropoxide (ViPr).However, they attributed the presence of the V 4+ peak to air exposure and not to any possible organic residues from the ViPr (despite having observed residual carbon by XPS).
The calculated composition analyses of the lms show that V 4+ accounts for a very small amount (less than 10% of total V), indicating that both thin lms are partially reduced if compared to the stoichiometric V 2 O 5 .Taking into account the atomic ratio of V and O (expected V : O ratio of 1 : 2.46 for V 2 O 5 ), we can be aware of the content of oxygen vacancies in the lms.A deviation from the stoichiometric V : O ratio, of 1 : 2.46, was observed for both lms, an indication of the presence of oxygen vacancies that arise from the reduction of V 2 O 5 , as expected. 72oreover, the peak at 533.2 eV of the O 1s is slightly higher in intensity for the lm made from the freshly prepared solution, and almost disappears in the thin lm made from the aged  Measurements were taken at 100 mW cm À2 AM1.5G.V 2 O 5 -IAP solution (as can be seen in Fig. 4a and S3 †).This is in good agreement with the replacement of the water molecules intercalated in V 2 O 5 by the IAP molecules in solution.Once prepared as a thin lm, the IAP evaporates from the V 2 O 5 layer leaving behind a thin lm without (or at least less amount) of water molecules.We can infer from these results that both thin lms are partially reduced: the lm prepared with fresh solution, by water, and the lm prepared with the 24 h-old solution, by IPA.
The optical band gap (BG), calculated from Tauc's formula, plot of a 2 E 2 against photo energy, 75 is shown in Fig. 4b.It reveals a slight difference in BGs between the thin lms fabricated from either fresh (red) or 24 h-old (green) V 2 O 5 -IPA solution, with values of 2.7 eV and 2.8 eV, respectively.The full He I scan of the ultraviolet photoelectron spectroscopy (UPS) analyses of the lms is shown in Fig. 4c.The work function (WF) values obtained were 5.15 eV (fresh) and 5.5 eV (24 h-old), respectively, as reected in the photoemission offset around 16 eV.These values are in good agreement with WF values of thin lms of V 2 O 5 fabricated in air. 15Finally, the values for the ionisation potential (IP), dened as the energy difference between the valence band (VB) edge and the vacuum level (E v ), are 7.66 eV (fresh) and 8.0 eV (24 h-old).
These results permitted the construction of the band energy diagram for both thin lms as shown in Fig. 5.To calculate the voltage of the OSC, we used the LUMO level of ZnO at 4.4 eV 57 and the HOMO level obtained experimentally for V 2 O 5 at 5.0 to 5.16 eV.The latter yields a V oc value of 0.56 V to 0.6 V, which is in good agreement with the experimental V oc values obtained for the OSCs shown in Fig. 3 (and very similar to the V oc values between 0.56 and 0.58 V observed in Fig. 1).However, we were unable to arrive at a clear conclusion regarding the V oc value of the solar cell that contained the V 2 O 5 thin lm made from the 24 h-old (green) V 2 O 5 -IPA solution (experimentally 0.38 V) since there is a wide range of possible reduction stages for V 2 O 5 that can be detected in IPA over time.Thus, based on the experimental and calculated values of V oc , we reasoned that the fabrication steps followed to obtain the V 2 O 5 thin lm affect the nal photovoltaic response of the OSC.Moreover, the nal V oc of the device is probably chiey dictated by the semiconductor oxide layers and by the HOMO/LUMO levels of the donor and acceptor materials of the active P3HT:PCBM layer.

OSCs with normal and inverted conguration with an Ag back metal electrode
One of the limitations for the fabrication of normal-conguration OSCs by low-temperature solution processing techniques is the requirement of low work function back metal electrodes, such as Al. 10,58While Ag electrodes can easily be printed from solution, there is currently no viable route for printing a stable Al electrode. 10This is a drawback that also limits the manufacture, by printing techniques, of tandem or multi-junction OSCs in the normal-conguration.Thus, in this section, we want to demonstrate that the fabrication of OSCs applying an Ag metal electrode is possible for both congurations when the V 2 O 5 HTL is applied.Fig. 6 shows the solar cells' energy band diagrams (a and b), the solar cell architectures in both congurations (c and d), and the corresponding IV curves and IPCE analyses for both types of devices (e and f).The photovoltaic parameters obtained are detailed in Table 3.The band energy diagrams in Fig. 6a and b are represented in relation to the relative energy levels of the acceptor (PCBM) and the donor (P3HT).The experimental values observed for the V 2 O 5 thin lm (5.1 eV) are very close to the energy level of the P3HT, and in good agreement with the work function of the Ag and the FTO electrodes responsible for the hole and electron collection respectively.Comparison of the photovoltaic response indicates a very similar behaviour, with V oc ranging between 0.54 V and 0.56 V and the FF between 47 and 48%.The main difference is observed on the J sc , which is lower for the OSC in the normal conguration in comparison with the inverted conguration.The difference in J sc also limits the PCEs, which is observed between 2.6% and 3% for the inverted conguration, and at around 2% for the normal conguration (see Table 3).This difference in PCE is further validated by the  corresponding IPCE responses: with 70% and 40% for inverted and the normal conguration, respectively.The dissimilarity in the performance between the two types of OSCs can be attributed to the greater light reection and the UV-lter effect imposed by the V 2 O 5 layer on the device.In the case where the device is illuminated from the FTO/V 2 O 5 side (see Fig. 6d and e), the V 2 O 5 layer could be acting as a UV-lter, limiting the amount of light reaching the cell.Adsorption spectra of the ZnO and the V 2 O 5 layers are shown in Fig. 6f.V 2 O 5 adsorbs at wavelengths up to 450 nm while in the inverted conguration (Fig. 6c), light enters the device from the FTO/ZnO side, where the ZnO layer blocks only the UV wavelength region below 380 nm.An interesting aspect observed is the value of V oc that is almost the same for both types of devices.This is an indication that the LUMO level of ZnO at 4.4 eV and the HOMO level of V 2 O 5 at 5.16 eV can be used to calculate the V oc of the normal conguration OSC. 57In the same way it was described before for the inverted OSC in Section 3.2.OSCs applying V 2 O 5 as the HTL 5,10,13,14,16-20,81-83 have been usually reported with an Ag metal electrode in the inverted conguration, 5,10,15,57 and an Al or Ca electrode in the normal conguration. 14,19,21,58In our work, the photovoltaic response of both types of OSCs seems to be independent of the Ag back metal electrode employed.This makes the OSCs amenable to fabrication by printing methods as the Ag metal electrode can simply be printed from solution. 5,10,44,79This also could be a step forward to the fabrication of more compatible recombination layers for TmOSCs. 10he selection of the adequate back metal electrode in OSCs has been the subject of extensive research work.The OSCs that have been studied to date contain only one oxide semiconductor used as the ETL (usually TiO 2 , TiO x or ZnO), and PEDOT:PSS as the HTL. 44,84Since the use of TMOs as both ETL and HTL is relatively new, we have not found any other work in which a high work-function metal electrode (e.g.Ag) is used for   The lifetime stability of the OSCs was analyzed under outdoor conditions for 1000 h.Fig. 7 shows the normalized PCE response observed with time for the inverted and normal conguration OSCs applying V 2 O 5 as the HTL.Initial results revealed that the light irradiation dose affects drastically the OSCs' response as can be observed in Fig. 7.A peak on PCE can be observed almost every time the light irradiation drops below 1 sun (100 W m À2 ), especially for irradiance between 0.6 and 0.8 suns (see also ESI, Fig. S4 and S5 †).Moreover, the lifetime analyses revealed better stability and longer lifetimes for OSCs with normal-conguration, staying at T 80 even aer 1000 h of analysis.The inverted-conguration OSCs revealed strong degradation reaching T 80 aer only 320 h of analysis.9][50]88 In our solar cells, we consider that the greater stability of the normal conguration OSC is partly due to the UV-lter effect that the V 2 O 5 layer can impose on the device when illuminated from the FTO/V 2 O 5 side, as already described.
In order to demonstrate that the OSCs would be more stable in the absence of UV light, we applied a UV lter to the inverted-conguration OSC.Two samples, one with the UV lter and the other without, were analysed outdoors under the same conditions.The lter (an adhesive UV lter lm that cuts UV light below 400 nm) was applied on top of the test cell.Fig. 7b shows the observed response for the rst 1000 h of analysis.The control sample performed just like the inverted OSC analysed in Fig. 7a, reaching T 60 at $500 h and T 40 at $1000 h.However, the sample with the UV lter remained at T 80 for many hours and was still stable aer $1000 h of testing.Thus we can demonstrate that elimination of UV light can improve the lifetime of the inverted-conguration OSC by several orders of magnitude.
In this work, we have demonstrated the high stability of OSCs containing V 2 O 5 $0.5H 2 O as the HTL despite the presence of water molecules in the layer.The degradation of the OSC lacking the UV lter indicates that V 2 O 5 is photoactive under UV light, and that the active P3HT:PCBM layer or the Ag electrode can interact with the V 2 O 5 HTL.Nevertheless, a UV lter is benecial and improves the OSC's stability.

Conclusions
In summary, we have demonstrated the rst example of stable organic solar cells (OSCs) containing a layered V 2 O 5 hydrate as the hole transport layer (HTL).V 2 O 5 is processed from a waterbased solution in air, resulting in a nal thin lm of formula V 2 O 5 $0.5H 2 O.The water molecules remain in the V 2 O 5 layered structure at temperatures below 250 C, which makes the thin lm highly stable under real working conditions.The HTL was employed in OSCs in either the normal or the inverted conguration, in which Ag was used for the back metal electrode.These types of OSCs can be fabricated totally by solution-processing printing in air, as they do not require the Al electrode found in normal-conguration OSCs.XPS, UPS and optical characterisation of the V 2 O 5 thin lms revealed differences based on the age of the V 2 O 5 -isopropanol (IPA) solution used for lm deposition.In lms made with a 24 h-old solution, reduction of the oxide (from V 5+ to V 4+ ) by IPA meant that subsequent re-oxidation (by photo-annealing) was required to achieve optimal photovoltaic performance.In contrast, the lms made with fresh V 2 O 5 -isopropanol solution directly exhibited peak performance and therefore did not require any photo-annealing.The normal-conguration OSCs do not require any photo-annealing because the V 2 O 5 thin lm is formed from an aqueous solution.Outdoor stability analyses of sealed OSCs containing V 2 O 5 as the HTL, in either the inverted or the normal conguration, revealed that the normal-conguration was highly stable.It remained at T 80 even aer 1000 h, probably due to the fact that it is illuminated from the FTO/V 2 O 5 side and to the UV-ltering effect of the V 2 O 5 layer.In contrast, the inverted-conguration OSC, which is illuminated from the FTO/ZnO side, was far less stable.Our hypothesis on the effects of the V 2 O 5 layer was corroborated by a subsequent test in which an inverted-conguration OSC, equipped with an external UV-lter, achieved comparable levels of stability to that of the normal-conguration OSC.

3. 1
Optimisation of the V 2 O 5 layer: concentration and layer thickness V 2 O 5 hydrate is obtained from an aqueous solution of sodium metavanadate (NaVO 3 ).During synthesis, the NaVO 3 dissolved in water is converted into metavanadic acid (HVO 3 ) via cationexchange.A condensation process over time results in the formation of decavanadates ([V 10 O 28 ] 6À ), and nally, a dark red solution is obtained, corresponding to vanadium pentoxide hydrate (V 2 O 5 $nH 2 O, where n varies).The variable water molecules in V 2 O 5 hydrate are partially eliminated when the V 2 O 5 $nH 2 O thin lm is formed on the glass/FTO substrate.The number of water molecules (n) in the formula V 2 O 5 $nH 2 O ranges from 0 to 2.2, depending on the annealing temperature: below 120 C, n ¼ 1.6; from 120 to 250 C, n ¼ 0.5; from 250 to 320 C, n ¼ 0.1; and annealing above 320 C promotes the total elimination of water (n ¼ 0) and the crystallisation of V 2 O 5 into its rhombic crystalline phase.64,65

Fig. 1
Fig. 1 Optimisation of the concentration of the V 2 O 5 hydrate solution used to create the hole transport layer in an inverted organic solar cell (glass/FTO/TiO 2 / P3HT:PCBM/V 2 O 5 /Ag).(a) PCE (%) and (b) J sc (mA cm À2 ).Measurements made at 100 W cm À2 AM1.5G.(c) V oc and (d) FF (%), (e) layer thickness vs. V 2 O 5 concentration and (f) AFM analyses of the V 2 O 5 thin film made with a concentration of 9 mg mL À1 .

Fig. 3
Fig. 3 Organic solar cells withV 2 O 5 hydrate as the hole transport layer in the inverted configuration (glass/FTO/TiO 2 /P3HT:PCBM/V 2 O 5 /Ag).Photovoltaic response of the cells fabricated from a freshly prepared (1) or a 24 h-old (2) V 2 O 5 -IPA solution.Using the fresh solution obviates the need for photo-activation of the device in air, as shown in the IV-curves and IPCE spectra from (2) to (3).Measurements were taken at 100 mW cm À2 AM1.5G.

Fig. 5
Fig. 5 Band diagrams for V 2 O 5 thin films obtained from freshly prepared (a) or 24 h-old and (b) V 2 O 5 -IPA solutions.E v : vacuum level; CB: conduction band; E f : Fermi level; and VB: valence band.

Fig. 6
Fig. 6 Schematic representation of the band energy diagram for the inverted (a) and normal (b) configuration of organic solar cells containing ZnO as the electron transport layer and water-based, solution-processed V 2 O 5 as the hole transport layer.The architecture of the inverted (e) and the normal (d) configuration OSCs.IV curves (c) and IPCE spectra (f) of the OSCs in each configuration.In both cases, an Ag back metal electrode was used.Measurements were taken at 100 mW cm À2 AM1.5G.
the normal conguration OSC.Greiner et al. recently described the effect of metal electrodes on the work function and band structure of MoO 3 at metal/metal oxide interfaces.The reduction of the oxide (from Mo 6+ to Mo 3+ ) in contact with the metal electrode results in a lower work function of the oxide, and the maximum value depends on the thickness of the oxide layer.85Hadipour et al. have employed an Ag metal electrode in different OSCs in the normal conguration, including the ones in which MoO 3 is the HTL.However, a thin layer of Ca between the active layer of P3HT:PCBM and the Ag metal electrode was employed for the normal conguration OSCs.86 Lidzey et al. have reported a study on different back metal electrodes in normal conguration OSCs in which MoO 3 is also the HTL.87The authors fabricated OSCs of the type ITO/MoO 3 / PCDTBT:PC 70 BM/metal electrode (note that no ETL was applied between the active layer and the metal electrode), using diverse, thermally evaporated metals (Ag, Al, Ca, Ca/Ag and Ca/Al).The nal photovoltaic performance of the solar cells was very similar in all cases, showing only slight differences among the devices.The authors chose the Ca/Al back electrode as the best one, owing to its slightly better photovoltaic response.Although their work involved only one TMO as the HTL (MoO 3 ) and did not entail the use of any ETL, it is the closest research work related to the one presented here by us (in terms of set-up and results).It also supports the idea that the photovoltaic performance of normal conguration OSCs containing metal oxides is probably independent of the back metal electrode used.Despite the advances made by Lidzey et al., our group and others, substantial studies are needed to clarify the role of the back electrode in these TMO-based OSCs.3.4 Outdoor stability analyses: normal vs. inverted OSCs, effects of UV lter

Table 3
Photovoltaic parameters of OSCs in the normal or the inverted configuration, with an Ag back metal electrode and water-based, solution-processed V 2 O 5 as the hole transport layer.Measurements were taken at 100 mW cm À2 AM1.5G a AE 1.1 47.20 AE 1.9 2.58 AE 0.2 AE 0.3 48.35 AE 2.3 2.10 AE 0.2 AE 0.01 10.69 AE 0.3 50.49AE 1.9 3.09 AE 0.1 a Average value from six samples.