2.1. Lead Iodide Layer Slot-Die Coating
Previous reports of slot-die coated lead iodide layers have made use of DMF as a solvent for formulations, this poses a challenge when moving to scaled-up production of perovskite solar cells on larger coating equipment as the solvent is highly toxic and would require complex expensive containment and extraction methods to safely use. DMSO has been utilised successfully as a solvent in many perovskite and perovskite precursor formulations and has successfully been used as an additive added in small quantities to single step formulations of perovskite inks for slot-die coating [30
] or for formulations of lead iodide in a 1–9 ratio with DMF used for slot-die coating of the lead iodide layer in sequential deposition, resulting in high performance devices [42
]. As well as being used as the sole solvent for slot-die coating lead iodide layers for use in sequential slot-die deposition (although not optimised for performance) [39
], it also has the benefit of being non-toxic. DMF has a Time Weighted Average (TWA) Workplace Exposure Limit (WEL) of just 5 ppm (from materials Safety Data Sheet (SDS)), much lower than other common solvents e.g., ethanol WEL TWA (1000 ppm) (from SDS), in comparison DMSO is not a hazardous substance according to Regulation (EC) No. 1272/2008e (from SDS) and so no exposure limits have been set for it. Although solvent containment methods would still be required when using such formulations, due to any toxic components that might be transported by the solvent vapours, removing the toxicity of the main solvent in the formulation is still critical to achieving an ink that can be used safely in a range of coating environments. Given this, the use of DMSO as the sole solvent for slot-die coating formulations of lead iodide is preferred and is the core development presented here.
The device structure used in this work includes a mesoporous titanium dioxide scaffold on to which the lead iodide layer is coated, this layer needs to fully and uniformly infiltrate the scaffold and to form a capping layer over the mesoporous titanium dioxide surface. As well as this, as noted in other works, to convert to perovskite efficiently the lead iodide must not form a too compact layer such that the subsequent MAI coating can not easily penetrate and convert the layer to perovskite [27
]. The morphology of the lead iodide layer formed will directly impact on the final perovskite morphology once converted, e.g., by the number of initial nucleation sites provided [60
]. To achieve adequate infiltration of the mesoporous layer it is necessary for any formulation to wet well on the surface of the layer.
The lead iodide ink for slot-die coating was formulated in solvent systems of either DMF or DMSO with solids content of 20 wt%. To determine wetting behaviour the contact angle of the DMSO formulation on a mesoporous titanium dioxide layer, as well as on the blocking layer, was measured and compared to that of the DMF formulation. The contact angles measured are summarised in Table 1
, the DMF-based ink was found to have a viscosity of 0.89 mPa·s compared to 2.54 mPa·s for the DMSO-based ink and surface tension of 35.3 m·Nm
compared to 40.3 m·Nm
for the DMSO-based ink. On the mesoporous surface the DMSO ink shows a higher initial contact angle than the DMF-based ink but both inks fully wet the surface over time, with static contact angles of <5
, with the DMSO-based ink wetting at a slower rate taking over 5 s to fully wet compared to around 0.5 s for the DMF-based ink. On the compact blocking layer surface the DMSO-based ink shows a relatively high static contact angle of 22.5
whereas the DMF-based ink shows complete wetting on the surface. The mesoporous layer was coated on the substrates as two stripes, with an underlying layer of spray coated blocking layer covering the entire substrate. The differences in contact angle of the two inks might be expected to result in different film qualities, as the initially coated wet film will spread depending on the relative wetting of the ink on the substrate.
To further investigate this, both lead iodide formulations were slot-die coated from an approximately 10
m lead iodide wet film thickness, to give films that once converted with MAI, would form approximately 600 nm thick perovskite dry films, with approximately 200 nm incorporated in the scaffold layer and 400 nm as a capping layer over the scaffold, as shown in Figure 2
. The coating machine has the substrate held on a platen that travels on a belt that moves the platen and substrate under the coating head and on into the oven unit. Coatings were made at 1 m·min
followed by directly travelling, over a distance of 30 cm, into the coater oven unit to be dried at approximately 105
C over a distance of approximately 30 cm with a line speed of 0.1 m·min
, to remove excess solvent, before being moved in to a fan oven and dried for a further 7 min at 100
The difference in rheology of the two inks result in markedly different film formation, images of both slot-die coating types which were then subsequently converted to perovskite by dip coating (to give improved contrast in images) are shown in Figure 3
, for the lead iodide films both dried at 105
C. The film produced using a DMF-based ink has over wet across the substrate and the stripe definition has been lost. This would be expected to be detrimental to device performance as the wet film thickness will vary and so the dry film thickness will no longer be as expected and will vary across the coating. The film produced using a DMSO-based ink shows the formation of the expected two stripe pattern with no over wetting, showing the improved film formation, compared to the DMF ink, under these coating and drying conditions. The improved film formation of the DMSO ink over the DMF ink can be explained by the poor wetting of the DMSO ink on the compact titanium dioxide surface that acts to contain the ink on the initially coated areas with underlying mesoporous scaffold, as well as by the relatively slower wetting of the DMSO-based ink on the mesoporous scaffold compared the DMF-based ink. Both films are continuous and do not show signs of coating defects such as ribbing or break up of the coating bead that would indicate a low flow limit being reached, this shows that despite the higher capillary number [61
] of the DMSO-based ink it is still suitable for slot-die coating under these coating conditions.
The morphology of the lead iodide layers was studied using scanning electron microscopy (SEM), images of the slot-die coated films, from either DMF or DMSO are given in Figure 4
. The film formed from the DMF-based formulation shows large needle like crystal grains, of the order of several microns in length, growing from the mesoporous titanium dioxide layer with voids between where the mesoporous layer is still visible. In comparison the film formed from the DMSO-based ink shows a more porous structure made up of smaller grains, with diameter in the order of a micron or less, there is also much less of the mesoporous scaffold visible. The morphology of the lead iodide film from the DMSO ink would be expected to react with MAI and convert to perovskite more rapidly than the DMF ink films due to the increased surface area of the smaller grains giving more reaction sites as well as the porous structure allowing the formulation to penetrate the film more easily, so speeding the reaction and conversion.
2.2. Methylammonium Iodide Slot-Die Coating
Having produced lead iodide films via slot-die coating the conversion to perovskite was investigated. The most commonly used method to convert lead iodide to the final perovskite is through the use of dip coating, normally by placing the lead iodide film in a dilute solution of MAI in 2-propanol (IPA), often for extended periods of time from a few minutes to tens of minutes. Slot-die coating of MAI formulations directly on to the lead iodide film has also been reported [27
], as the solvent rapidly evaporates any reaction to form perovskite occurs within a much shorter time period than for most dip coatings so achieving complete conversion to perovskite in this time frame is challenging. Heating of the substrate has been reported to improve the conversion to perovskite [27
] by accelerating the reaction (but also speeds evaporation of solvent), but this adds extra complications to the coating process so ideally a coating method avoiding this would be preferred.
To determine the conversion to perovskite and to test and compare the performance of devices, incorporating the slot-die coated DMF and DMSO-based lead iodide films, dip coating was used initially. The films were placed in a solution of MAI for 30 min, a method previously found to give good conversion to perovskite.
X-ray diffraction spectra of the perovskites formed from both films are given in Figure 5
, in both cases the expected peaks for the perovskite structure are present, with the peak at 14.1
dominant, representing the 110 reflection, for both films there is no signal for lead iodide at around 12.7
. This indicates good conversion to the final perovskite, using dip coating, for films produced from both lead iodide inks.
As discussed previously in the context of wetting, SEM images of the perovskite formed are given in Figure 4
, in both cases very similar structures are formed with excellent coverage over the underlying mesoporous layer and crystal grains of around 1 micron diameter formed.
gives the median current-density voltage (JV) curve parameters of devices masked to 0.09 cm
made using the DMF (structure A) or DMSO (structure B) lead iodide films converted using dip coating, in both cases high efficiency devices are produced, results are also summarised in Figure 6
as box-plots. The DMSO-based devices show greater power conversion efficiency (PCE) than the DMF-based devices, this is mostly attributed to an improved light shunt resistance (Rsh), of 1249 compared to 1017 ohms·cm
(taken from the reverse light JV scan curve), so improved fill factor (FF), open circuit voltage (Voc) and short-circuit current-density (Jsc). This could be attributed to the better stripe coating quality and more uniform lead iodide coating of the DMSO-based ink that gives a more optimised perovskite film thickness that results in greater charge carrier generation and improved shunt resistance and fewer losses of charge carriers to recombination.
Having established that the DMSO-based lead iodide films could form good quality perovskite layers and high performance devices the use of slot-die coating, rather than the conventional dip coating, for the deposition of the MAI ink was investigated. To achieve complete conversion of lead iodide to perovskite the MAI formulation would have to sufficiently wet and infiltrate into the lead iodide film to bring the two components in contact with each other and to react and form perovskite [62
]. For the dip coating process the use of other solvents has been investigated and in some cases found to improve conversion to perovskite, ethanol has been shown to be superior to IPA for the dip coating process, where it was suggested the lower viscosity of ethanol improved the kinetics of the reaction [64
As well as converting to a high degree the morphology of the resulting perovskite film is critical to device performance and the carrier solvent for MAI will strongly influence the morphology formed. Alcohol solvents with various structures have been shown to influence the morphology of sequentially grown perovskite crystals, in particular a dissolution–recrystallisation process (Ostwald ripening effect) dependant on the molecular polarity of the solvent has been suggested to be influential [65
]. The static relative permittivity, used as a macroscopic measure of the polarity of the solvent, has also been highlighted as important for sequentially deposited perovskite crystal growth, the size of grains and the number of grain boundaries formed in films, with a higher permittivity facilitating the formation of films with larger grains and fewer grain boundaries [66
As well as this, the rheology of the solvent and in particular viscosity will determine how rapidly the formulation infiltrates the porous lead iodide film [67
]. For slot-die coating the volatility of the solvent is also expected to be critical, as the ink wet film is generally only a few tens of microns and so solvent will quickly evaporate giving a short time frame for both the formulation to infiltrate the film and for the reaction to occur [68
]. The solubility of MAI in the solvent will also be important as MAI will rapidly crystallise out of solution from weak solvents as the solvent evaporates. The MAI will no longer be available to react, potentially crystallising on the surface of the film resulting in a poor interface with the hole transport layer deposited on top of it [69
] and forming large crystallites providing a pathway for degradation reactions, for instance with the top electrode or via atmospheric moisture. The overall macroscopic coating quality of the layer including film defects such as discontinuity will depend on the rheology of the formulations and the particular coating conditions.
In order to investigate the role the solvent in conversion to perovskite for slot-die coating, formulations of MAI in several different alcohols, with various rheologies and volatilities, were prepared, these are listed in Table 3
. Methanol has a low viscosity but is also very volatile, ethanol has a slightly higher viscosity and is less volatile, IPA (the solvent used for most slot-die sequential depositions reported) is of similar volatility to ethanol but has a higher viscosity and 1-butanol has a higher viscosity still and is less volatile.
It should be noted that the concentration of MAI in the formulation can also alter the crystallisation process and the morphology of perovskite films formed [70
]. To control for this the concentration of MAI in the formulations was kept constant and was chosen as a balance of being accessible within the solubility limits of the solvents and not requiring an excessively great wet film thickness to achieve a 1:1 molar ratio of MAI and lead iodide. Too great a wet film thickness could cause non-uniform drying or cause a build up of ink at the upstream lip of the coating head e.g., flooding/dripping and a loss of pre-metering [71
] when coating at too low a speed.
All the MAI formulations showed complete wetting on the lead iodide surface with static contact angles less than 5. The MAI formulations were slot-die coated onto lead iodide coated substrates in order to establish the conversion to perovskite for each ink and the morphology of the layer formed. For all the formulations the coating and drying conditions were kept constant, with a coating speed of 1 m·min, followed by allowing excess solvent to evaporate at room temperature and then drying in a fan oven at 90 C.
shows the XRD spectra taken for each of the films, clearly the level of conversion to perovskite is low for the methanol (structure D) and 1-butanol (structure F)-based formulations, indicated by the large peak at 12.7
corresponding to residual lead iodide. The poor conversion for the methanol-based ink films can be linked to the high volatility of the solvent and the rapid evaporation of it from the film resulting in there being little time for more extensive conversion to perovskite. The poor conversion of the 1-butanol ink films can be linked to the rheology of the ink with the higher viscosity slowing the rate of reaction and penetration of solution into the lead iodide film. The IPA (structure C) and ethanol (structure E)-based formulations show slightly greater conversion to perovskite, with the ethanol slightly greater, but not complete conversion as seen for the dip coated films (structure B).
shows SEM images of the films formed from each formulation, the methanol-based formulation films show what appear to be large perovskite crystals growing out of a bed of lead iodide with very limited and sporadic conversion to perovskite. The high volatility of the solvent and the rapid evaporation of it from the film results in little time for conversion to perovskite to take place and this is localised to small areas. The film also appears to have lost some of the initial lead iodide film structure with the lead iodide recrystallising to a more planar structure, which might be due to the high relative permittivity of methanol facilitating the dissolution–recrystallisation process. The IPA based formulation results in films with loosely packed crystal grains of the order of hundreds of nanometres in diameter with a general structure similar to that of the initial lead iodide film. Whereas the ethanol-based formulation shows smaller more densely packed crystals, where the surface is much more uniformly structured, possibly due to the higher static relative permittivity of the solvent increasing the dissolution-crystallisation of the initial film and loss of initial structure. The 1-butanol formulation results in small grains but more loosely packed than the ethanol formulation, Clearly a balance of volatility, rheology, solubility of MAI and permittivity of the alcohols plays a crucial role in the conversion to perovskite and the morphology formed.
Photovoltaic devices were fabricated using the different slot-die coated MAI films, Table 2
and Figure 6
summarise the JV scan photovoltaic parameters measured for these. Clearly the methanol-based formulation (structure D) results in poor device performance, as would be expected from the poor conversion to perovskite. The devices produced from the IPA and 1-butanol-based formulations (structure C and F) showed some cells with better efficiencies than those made using the methanol formulation, but not as good as for dip coated films, with a low short-circuit current density being the main cause of poor performance. The 1-butanol-based formulation devices also show a large spread in device performances, despite the very poor overall level of conversion inferred from the XRD results this suggests some areas of more optimal conversion to perovskite and some less. The ethanol (structures E) devices showed some very encouraging performances, with power conversion efficiencies approaching that of the dip coated cells but with a wide spread in the performances measured. This despite the relatively poor level of conversion to perovskite, but excess lead iodide even to levels of 20% or more has been shown to result in working devices [69
]. The IPA, methanol and 1-butanol-based slot-die MAI ink devices suffered from increased series resistance (taken from the reverse scan light JV curve) compared to the dip coated films, of 10.8, 20.1 and 9.3 ohms·cm
respectively compared to 7.0 ohms·cm
for the dip coated films and 6.6 ohms·cm
for the ethanol-based slot-die ink, which could be due to the lower conversion to perovskite and residual resistive lead iodide. All of the devices show significant hysteresis in the JV curves between reverse and forward scans.
In an attempt to further improve the device performance a modified lead iodide coating process was developed to produce more labile lead iodide films in order to increase conversion to perovskite when slot-die coated with the MAI in ethanol ink. The substrate was pre-heated to 100
C and then the lead iodide ink coated on to the hot substrate. This resulted in lead iodide films that were found to convert to perovskite almost fully when MAI in ethanol was slot-die coated over them, as shown in the XRD spectra in Figure 7
(structure G). The surface morphology of the resulting films is shown in the SEM images in Figure 9
, the surface shows more undulations than that produced on room temperature substrate coated lead iodide films, with more small areas of the mesoporous scaffold visible, but still with good overall coverage of the scaffold. Comparing the perovskite films formed from the room temperature and heated substrate lead iodide films with slot-die coated MAI from the ethanol formulation the grain size is increased for the heated substrate, but the film is less planar and seems to retain more of the structure of the initial lead iodide film.
Performance of devices made using these films are given in Table 2
and Figure 6
, (structure G). The average PCE achieved is significantly improved compared to the cells with lead iodide coated at room temperature and slot-die coated MAI in ethanol formulation (structure E). This is mainly due to an increase in Jsc, which can be related to the improved conversion to perovskite resulting in more material contributing to photo-current generation. There are also smaller improvements in Voc and FF that also contribute to the overall improved median PCE of 11.0% compared to 8.0% for structure E, which is comparable to the efficiency (10.7%) of the dip coated devices (structure B). The shelf-life stability of devices using slot-die coated films of lead iodide coated on heated substrate with either dip coated or slot-die coated MAI are compared. Devices were encapsulated with Kapton tape, UV curable adhesive and a glass coverslip and stored in the dark in a sealed box containing desiccant, devices were then tested again after 442 days. The slot-die coated devices showed similar levels of degradation to the dip coated devices and retained approximately 60–70% of the original efficiency values, in line with what is usually seen for this device structure and perovskite type [32
], with an increase in series resistance and decrease in Jsc, but little change in FF, as shown in Table 4
. This result suggests that the slot-die coating process developed for MAI is not detrimental to shelf-life stability compared to dip coating.