Development of Greener and Stable Inkjet‐Printable Perovskite Precursor Inks for All‐Printed Annealing‐Free Perovskite Solar Mini‐Modules Manufacturing

Inkjet‐printing is considered an emerging manufacturing process for developing perovskite solar cells (PSCs) with low material wastes and high production throughput. Up‐to‐now, all case studies on inkjet‐printed PSCs are based on the exploitation of toxic solvents and/or high‐molarity perovskite precursor inks that are known to enable the development of high‐efficiency photovoltaics (PVs). The present study provides a new insight for developing lower‐toxicity, high performance and stable (for more than 2 months) inkjet‐printable perovskite precursor inks for fully ambient air processed PSCs. Using an ink composed of a green low vapor pressure noncoordinating solvent and only 0.8 m of perovskite precursors, the feasibility of fabricating high‐quality and with minimum coffee‐ring defects, annealing‐free perovskite absorbent layers under ambient atmosphere is demonstrated. Noteworthily, the PSCs fabricated using the industry‐compatible carbon‐based hole transport material free architecture and the proposed ink present an efficiency >13% that is considered on the performance records for the under‐consideration PV architecture employing an inkjet‐printed active layer. Outstanding is also found the stability of the devices under the conditions determined by the ISOS‐D‐1 protocol (T95 = 1000 h). Finally, the perspective of upscaling PSCs to the mini‐module level (100 cm2 aperture area) is demonstrated, with the upscaling losses to be as low as 8.3%rel dec−1 per upscaled active area.


Introduction
Since their first report, perovskite solar cells (PSCs) have demonstrated an unrivaled development, becoming a gamechanger in light-to-electricity conversion, thus getting tremendous attention in the photovoltaic (PV) field.Today, just a decade later, the perovskite research community has already achieved major strides toward increasing the performance of these devices, surpassing the efficiency of most of the well-established PV technologies such as crystalline silicon, cadmium telluride and copper indium gallium di-selenide.Considering that the power conversion efficiency (PCE) of PSCs has now reached to 25.7%, [1] the current hot challenges in the field are to attain high operational stability and the effective upscaling of the technology at the standard module level utilizing industrial-compatible and green manufacturing. [2]iming at a low levelized cost of electricity (LCOE) and solar cell attributes that would allow perovskites for real market competition to the commercialized PV technologies, new sustainable development options are continuously under investigation.Regarding the small scale, new PV materials and device architectures are intensively studied, with the focus now being set on low-cost and complexity systems.For the transition of the technology to the commercialization level, simplified-structured PSCs (electron/hole transport material free solar cell designs) take today a leading position in research in replacement to the conventional n-i-p and p-i-n noble-metal back-contact structures. [3]Among them, the carbonbased hole transport material free (C-based HTM-free) architecture is considered the frontrunner to the PV market, mainly stemming from the inexpensive carbon materials, the full compatibility with printing techniques and the high operational stability that usually characterize these devices. [4,5]On the other hand, as a quite important aspect of lab-to-fab transition, ambient air-processing has been increasingly applied in the last years, where new manufacturing methods and strategies that can allow the high-quality development of PSCs under high-humidity conditions are demonstrated. [6]Considering the intended environmental footprint of these devices, the recent research trend also proceeds to green chemistry manufacturing approaches that would achieve the reduction of the global warming potential and cumulative energy demand of this technology. [7]Herein, a special focus is set on the perovskite precursor solution, aiming to reduce the toxicity of this system, which mainly arises from the entity of lead compounds and toxic solvents.Today, effective strategies have demonstrated the development of high performance double perovskites or lead-free perovskites using partial or complete substitution of the lead ion, while biofriendly approaches such as green solvent engineering are also increasingly adopted.Moreover, to promote the commercialization and volume production of perovskite PVs, the development of highquality annealing-free perovskites is applied as an extra effective strategy to lower the capital budget of the equipment footprint of these devices.To realize high-quality perovskite crystallization in this direction, solvent and additive engineering, as well as various post-treatment processes, have already been proposed and developed. [8,9]egarding the upscaling of PSC technology, several alternative manufacturing processes have already been demonstrated with success for the fabrication of PVs on a large scale, including slot-die coating, screen-printing, spray-coating and inkjetprinting. [10]Among them, the manufacturing way of piezoelectric drop-on-demand inkjet-printing provides important attributes such as high manufacturing control, low material wastes and high throughput. [11]Using inkjet-printing, breakthrough achievements have been demonstrated in the field, with one of the most important being the fabrication of the largest certified perovskite solar module by Panasonic, which attained an efficiency on the order of 18% on an active area of about 800 cm 2 . [10]evertheless, continuous efforts are also put toward the maturation of the inkjet-printing processing of PSCs, with the key inkjet challenges to be thoroughly considered in case studies devoted to provide sustainable solutions to the faced manufacturing issues, with one of the most critical ones being the coffee-ring effect. [12,13]o these challenges, several critical steps should be yet taken to proceed with a real high-quality and throughput industrial manufacturing of PSCs.
Considering all these struggles in the field, the present work aims to introduce a lower toxicity, high performance and stable perovskite precursor ink for inkjet-printing processing of PSCs employing for the first time a solution system composed of gamma-valerolactone (GVL) as a green solvent and lowconcentration of perovskite precursors (0.8 m).The new ink is evaluated in terms of its jettability, chemistry and wettability on the electron transport material (ETM).On the other hand, the characteristics of the developed perovskite absorbent layers are thoroughly investigated from the structural, chemical, morphological and optical point of view, while subsequently, solar cell devices are also developed and characterized in terms of both their performance and stability.In all cases, results arising from the usage of a reference 2.25 times higher molarity dimethylformamide (DMF) based perovskite precursor ink are also provided for comparison purposes (DMF is one of the most commonly applied solvents for the preparation of perovskite precursor solutions, including inkjet-printing process).Finally, the perspective of scaling up the PSCs to the mini-module level (100 cm 2 aperture area) utilizing the proposed ink and a C-based HTM-free device architecture is demonstrated.

Jettability of the Perovskite Precursor Inks and Wettability of the Electron Transport Material by the Inks
For the successful development of an inkjet-printed device, the applied ink should at first meet the requirements of the printing process, which are determined by the specific machinery/equipment that is utilized.Among the most critical characteristics that should be considered is the jettability of an ink.In a typical development procedure, one should first formulate the ink according to the specification window of the equipment.Weber (We), Reynolds (Re) and Ohnesorge (Oh) numbers are some of the parameters that should be calculated to assess the printability of an ink.The values of We, Re and Oh numbers can be estimated using Equations (1-3), where  is the density of the ink, d is the diameter of the nozzles (herein d = 17 μm), v is the jetting velocity,  is the surface tension of the ink and  is the viscosity of the ink. [14] = dv 2 ∕ (1) For a typical inkjet-printer employing a printhead of 20 μm nozzles diameter, the viscosity of the ink should be between 1 and 25 mPas and the surface tension of the ink between 25 and 50 mNm −1 . [15]The flash point of the solvent of the ink should also be higher than the printing temperature to avoid nozzles clogging and early drying.The ejection speed of the ink can be set after considering a minimum value computed from the We number, where a We value higher than 4 is recommended.For optimum jetting, the ratio of inertia to the viscosity force (i.e., Re number) should be higher than 1, and the value of Oh number has to range between 0.1 and 1 to achieve stable droplets ejection and without satellites.Finally, a We 1/2 Re 1/4 value lower than 50 is required to avoid droplets splashing. [14]In the present case, the physical properties and the calculated parameters related to the jettability of the reference (1.8 m DMF-based ink) and the proposed ink (0.8 m GVL-based ink) at printing temperature are tabulated in Table 1.As it is perceived, in both cases, the calculated values meet the rheology property limitations for their inkjet-printing (jetting velocity is set at 6 ms −1 ).Following up, the wettability of the substrate by the ink is also critical and should be considered to achieve a high-quality inkjetprinted film.This is even more important to be investigated in developing PSCs since the wetting of a solar cell material by the perovskite precursor ink has a significant effect on the heterogeneous nucleation of the perovskite.Better wetting implies that perovskite nuclei and substrate active sites have a better affinity, considerably lowering the nucleation barrier owing to the effective decrease of interface energy. [16,17]In this way, a high nuclei density and pinhole-free perovskite absorbent layer can be developed.Nevertheless, considering the printing processing, over-wetting (contact angles lower than 5°) should be avoided to obtain optimum printing performance. [18]Herein, the so-called "surface-wetting" is determined by contact angle measurements (sessile drop goniometry).Considering that the ETM is porous, the initial contact angle was measured to evaluate the wettability.This is because the static contact angle is usually determined almost at zero degrees, attributed to the droplet infiltration into the pores network of the substrate.For picoliter drops, the initial contact angle used for the wettability measurements is usually taken after 2 ms (time zero is when the droplet contacts the substrate) in order to avoid isolation damping. [19]Herein, the image showing the initial contact angle is captured at t = 10 ms (see Figure S1, Supporting Information).As it is perceived, the wettability of the ETM by the perovskite precursor inks is good for both types of inks, with the smaller contact angle to be obtained for the 0.8 m GVL-based ink.

Characteristics of the Perovskite Absorbent Layers
X-ray diffraction (XRD) analysis was carried out to investigate the crystal quality of the inkjet-printed perovskite absorbent layer developed on the top of the conductive substrate and ETM by the use of the 1.8 m DMF-based and 0.8 m GVL-based perovskite precursor inks.Except for 20 min annealing applied for the reference samples, annealing-free perovskite absorbent layers were also evaluated regarding their crystallinity, for comparison purposes.The XRD patterns of the four different samples are presented in Figure S2 (Supporting Information), while the corresponding graphs for the best case of annealing duration for the different ink-processed samples are presented in Figure 1.Additionally, a comparison of the crystallinity of the perovskite absorbent layer developed by the use of a fresh and 2 months stored/aged 0.8 m GVL-based ink took place and the corresponding XRD patterns are presented in Figure S3 (Supporting Information).The full width at half maximum (FWHM) calculated at the main crystal planes of CH 3 NH 3 PbI 3 (i.e., 110, 220 and 310) for all samples are tabulated in Table 2.
The fabrication of high-quality perovskite crystals is important to achieve a high light-to-electricity conversion efficiency.Usually, in inkjet-printing, high molarity perovskite precursor inks are applied to develop high crystalline and homogeneous antisolventfree perovskite absorbent layers under ambient atmosphere. [13]his is because a high molarity perovskite precursor ink can favor the development of homogeneous and large crystals, without discontinuities in the crystal structure of the layer.However, this is not one size fits all case observation.Herein, of particular interest is that the crystal characteristics of the perovskite films developed by the use of 1.8 m DMF-based ink and 0.8 m GVLbased ink did not differ intensively even though the perovskite precursors concentration in the former ink is 2.25 times higher than the latter one.This can be attributed to the following reasons.At first, the viscosity of the latter ink is much lower than the former one.In this way, the wetting of the substrate by the ink is much better, as is also demonstrated by the sessile drop goniometry, which, in turn, favors the heterogeneous nucleation of the perovskite.Besides, GVL solvent combines two attributes that can lead to the development of high-quality perovskite films through a single-step antisolvent-free process.These attributes are (1) the negligible coordination capability of the solvent with the perovskite precursors and (2) the relatively low vapor pressure that this solvent demonstrates.
In contrast to the DMF that acts as a Lewis base and can coordinate easily with Pb 2+ ions to form Lewis base-acid intermediate adducts, in the case of GVL-based ink, there are no/negligible solvent-solute coordination complexes (see Fourier transform infrared (FTIR) spectroscopy analysis, Figure S4a, Supporting Information). [22,23]The development of these complexes in the ink hinders the easy removal of the solvent during the crystallization process, and even if the solvent is volatile, the morphology of the perovskite films will be uneven/discontinuous if an extra manufacturing step, such as antisolvent extraction, gasquenching or vacuum-assisted drying, is not applied. [24]On the other hand, the relatively low vapor pressure of GVL at printing temperature (about 15 torr at 50 °C) ensures manufacturing quality, where the issues of poor crystallized films and their poor contact with the substrate appearing in a corresponding case of usage of high-volatile noncoordinating solvents are avoided. [25,26]Even though the usage of a high-volatile solvent can highly reduce the time frame toward supersaturation, the crystal growth is not favored under high solvent evaporation rates, subsequently leading to small grains and discontinuous perovskite crystals. [27]To deal with these issues, more complex systems containing mixtures of solvents or additives should be employed to achieve a balance between nucleation and growth rates, rendering high-density nuclei, templating their uniform growth, and consequently developing a homogeneous and compact active layer composed of highquality perovskite crystals. [26,28]e negligible coordination capability of GVL with the perovskite precursors can also be verified by the development of a high amount of PbI 3 − iodoplumbate complexes in a fresh GVL-based ink as perceived by the ultraviolet-visible (UV-VIS) absorbance analysis (see Figure S4b, Supporting Information), something that was not observed in the case of the DMF-based ink.The development of PbI 3 − and PbI 4 2− complexes is a timedependent process, and it is highly related to the coordination capability of the applied solvent with the solutes, where the stronger the interaction between solvent and PbI 2, the lower number of coordinated I − ions in the iodoplumbate complexes that appear. [29]odoplumbate complexes containing a higher content of coordinated I − in the precursor solution with weak coordination solvents or additives are requisite for the development of perovskite crystals.In contrast, strong coordination solvents or additives reduce the number of coordinated I − in iodoplumbate complexes, suppressing the perovskite formation. [24]This could also be the main reason behind the observation that the perovskite absorbent layers developed by the use of the GVL-based ink crystallize easily and present high-quality crystals even in the case of an annealingfree process.Nuclear magnetic resonance (NMR) analysis of perovskite powders scratched from inkjet-printed perovskite films, developed using either DMF or GVL as a solvent, was also conducted to identify possible solvent remains in annealed and annealing-free perovskite films.As shown in Figure S5 (Supporting Information), in the case of the samples that arose from the inkjet-printing of DMF-based ink without the subsequent annealing of the film, a small amount of DMF solvent was identified, verified by the chemical shifts of peaks at 7.94, 2.88 and 2.73 ppm.On the contrary, no chemical shifts attributed to DMF protons appeared for the 20 min annealed powder.Interestingly, no chemical shifts attributed to the GVL solvent were also detected in the samples that arose from the inkjet-printing of GVL-based ink without the subsequent annealing of the film.This means that even if the annealing process after the printing of GVL-based ink is skipped, a complete conversion of the precursor materials to perovskite structures without the presence of GVL residuals can be attained.
Regarding the stability of the perovskite precursor inks, the coordination capability of the solvent with the perovskite precursors is also crucial to attain a stable ink in time.The coordination between lead halides and solvent molecules can lead to the formation of precipitates in the perovskite precursor ink after a period of storage.Even if the amount of these precipitates is low, they highly influence the stoichiometric compositions of the system, which subsequently results in a lower performance when applied to develop PV devices. [30]Herein, due to the strong coordination of DMF with the perovskite precursors, as well as the high molarity of perovskite precursor ink, the development of a great amount of CH 3 NH 3 I-PbI 2 -DMF precipitates appeared even after a few days of its storage.
To give further evidence on the characteristics of the perovskite absorbent layers fabricated by inkjet-printing the 1.8 m DMFbased and 0.8 m GVL-based perovskite precursor inks, the samples were characterized by field emission scanning electron microscopy (FE-SEM).The top-view and cross-sectional FE-SEM images of the two different samples fabricated after an initial optimization of the printing parameters to achieve the highest light-to-electricity conversion by the PV devices are presented in Figure 2. As is observed in Figure 2a,c and clearer view in the higher magnification top-view FE-SEM images presented in the Figure S7a (Supporting Information), the usage of the 1.8 m DMF-based ink produces uneven, rod-shaped perovskite structures, resulting in incomplete surface coverage of the underneath ETM.Besides, with a closer look at the low-magnification topview FE-SEM images showing the morphology of the film at a larger area, it can be perceived that the film presents line-pattern defects, with low-density rod-shaped perovskite structures to be developed periodically.This can also be observed in the top-view optical microscopic image captured by the built-in camera of the inkjet-printer (see Figure S8a, Supporting Information).On the other side, the usage of the 0.8 m GVL-based ink results in a uniform and compact perovskite film (see Figure 2b,d).In this case, the film is composed of large perovskite grains, without presenting pinholes and caves in their structure, as well as the line-pattern defects that appeared in the case of the DMF-based ink (see Figures S7b and S8b, Supporting Information).
The development of low-quality antisolvent-free perovskite absorbent layers when using perovskite precursor inks prepared by coordinating solvents (e.g., DMF) is commonly reported in the literature.These rod-shaped perovskite structures are emanated from the nonfully coordination of the CH 3 NH 3 I molecules with the PbI 2 , even in the case of inks composed of CH 3 NH 3 I:PbI 2 at 1:1 molar ratio, arising from the solvent-PbI 2 coordination. [24]In this way, (CH 3 NH 3 ) 2 (DMF) 2 Pb 3 I 8 and (CH 3 NH 3 ) 2 (DMF) 2 Pb 2 I 6 phases are the dominant intermediate phases in the perovskite precursor ink, both containing facesharing PbI 6 octahedral, which presents high anisotropy in the surface energy of perovskites growth, resulting to the formation of these perovskite rod-shaped structures.On the other side, the usage of a perovskite precursor ink composed of a low-volatility weak/noncoordinating solvent (e.g., GVL) presents the following important attributes.At first, there is better control of the stoichiometric coordination of the perovskite precursors, while also the development of intermediate phases that retard the crystallization process and lead to the formation of uneven morphology films is eliminated.By controlling the speed of solvent evaporation (not too fast), there is also better contact of the film with the substrate, while the anisotropic growth of perovskite crystals faced in the case of usage of high-volatility noncoordinating solvents is avoided (i.e., achieving a balance between nucleation and growth rate).
Talking about the periodical morphological defects that appear in the perovskite films developed by the 1.8 m DMF-based ink, these arise from the coffee-ring effect and Marangoni convection evolved during the inkjet-printing procedure, as is also demonstrated in previous investigations. [13,31]According to these reports, the different evaporation rates of the solvent in a drop surface induce high capillary flows, while the surface tension gradient from the regions of high material concentration to the regions of low material concentration produces Marangoni convection, finally leading to the development of a zone of low material density (called "depletion zone").Herein, the evolution of the coffee-ring effect for both cases of inks can be viewed in the drying drop experiments demonstrated in Video S1 (Supporting Information).As it is perceived, in both scenarios, the crystallization of the material begins near the drop edge, where the evaporation of the solvent is strong, followed by an outward convective flow.However, as indicated by comparing the crystallization kinetics of the two cases, the fluid dynamics are significantly more prominent for the wet film produced by the 1.8 m According to previous studies, this phenomenon is much more pronounced in the case of inks composed of low-concentration perovskite precursors. [13]On the other side, by the usage of GVLbased ink, a better wetting of the ETM by the ink is achieved, while the development of perovskite crystals is found to be totally uniform, favored under the tailored evaporation of the GVL solvent.At this point, it is noteworthy to mention that the vapor pressure of GVL solvent presents a low variation with temperature (ranging from 0.65 kPa at 25 °C to 3.5 kPa at 80 °C), unlike the commonly used solvents in PSCs development, such as DMF. [25,32]This is an important attribute that is highly appreciated in industrial manufacturing, offering a wide range of temperature processing window.
To shed more light on the differences in the characteristics of the perovskite films fabricated by the usage of the 1.8 m DMFbased and 0.8 m GVL-based inks, optical measurements were conducted focused on the visible spectrum of light, and the results are presented in Figure 3.To this aim, the light-harvesting efficiency (LHE) of the two different samples was also calculated using Equation (4), where A is their absorbance spectrum and R is their reflectance spectrum.Additionally, the optical bandgap of the perovskite films was determined considering Tauc's plots presented in Figure S9 (Supporting Information), assuming a direct transition.
As it is observed in Figure 3, the two samples present almost the same LHE, with the values ranging on the order of 90% for almost the whole spectrum of visible light.Namely, although the morphology of the perovskite absorbent layers fabricated by the usage of the 1.8 m DMF-based and 0.8 m GVL-based inks presents significant differences, the optical analysis showed negligible alteration in their optical characteristics.Small was also the difference in the values of the optical bandgap of the compared perovskite films, determined at 1.55 and 1.57 eV for the case of usage of the DMF-based and GVL-based inks, respectively.These values are found in line with the ones usually reported in the literature for CH 3 NH 3 PbI 3 structures. [33]Small variations in the optical bandgap of perovskites can be obtained for several reasons, including the development of structures from nonstoichiometric coordination of the perovskite precursors and the existence of uncoordinated PbI 2 in the perovskite film. [34,35]This can also be affected by the type of solvent (coordinating or not) applied to develop the perovskite precursor solution.

Characteristics of the Perovskite Solar Cells and Mini-Module
To evaluate the suitability of the GVL-based ink application for the development of all-printed PSCs, small-sized PV devices were initially fabricated and characterized, and subsequently compared to corresponding devices developed by the reference DMF-based ink.Besides, to investigate the potential of decreasing the manufacturing footprint of PSCs, solar cells employing annealing-free perovskite absorbent layers developed by the two different inks were also evaluated regarding their light-toelectricity conversion efficiency (see Table S1, Supporting Information).Figure 4 presents the current-density-voltage (J-V) curves of the most efficient PSCs fabricated by the 1.8 m DMFbased and 0.8 m GVL-based inks, while the mean value and standard deviation of the electrical characteristics of the corresponding solar cells collected on five identical devices are tabulated in Table 3.
As it is observed, the PSCs developed by the 0.8 m GVL-based ink present a quite satisfactory light-to-electricity conversion The mean values and standard deviation have arisen from the characterization of five identical devices.
efficiency, which is found to be even higher than the efficiency attained by the reference devices developed using the 2.25 times higher molarity DMF-based ink.More specifically, by inkjetprinting the 0.8 m GVL-based ink, the fabrication of annealingfree PSCs of 13.07%PCE was attained.Noteworthily, this value of performance is considered one of the highest reported for perovskite PV devices developed on the basis of C-based HTM-free architecture employing an inkjet-printed perovskite absorbent layer (see Table S2, Supporting Information), while it is the only case of report (according to the authors' knowledge) on an inkjetprinted PSC processed without annealing of the perovskite absorbent layer after the printing process.By comparing the electrical characteristics of the solar cells presented in Table 3, it can be perceived that the increased performance of the PV devices developed by the GVL-based ink compared to the devices developed by the DMF-based ink, is mainly attributed to the increased opencircuit voltage (V OC ), while a small increase was also observed in the fill factor (FF) value.On the other hand, the annealing-free device developed by the DMF-based ink was found to present a lower performance than the one of the reference 20 min annealed perovskite PV device, something that was not the case for the devices developed by the GVL-based ink, in agreement with the XRD analysis.The fabrication of high-efficiency PSCs without the need for postannealing processing of the inkjet-printed perovskite absorbent layer is highly appreciated in industrial manufacturing since, in this way, the energy consumption and the production cost of the devices can be further reduced.A more and in-depth discussion on the origins of the observed differences in the PV characteristics of the compared solar cells is provided on the basis of the quantum efficiency and the impedance characteristics, as followed in the next paragraphs.Finally, an important result derived from the present investigation is that the GVLbased ink could be applied in the fabrication of high-efficiency PSC devices even after a period of its 2-month storage/aging (see Table S3, Supporting Information), underlying the development of an industry-compatible inkjet-printable ink considering its high stability and low toxicity.As a critical challenge faced in the PSC technology, which highly affects the performance and stability of this type of PV devices, the hysteretic behavior of the developed solar cells was also determined. [36]Herein, the hysteresis index (HI) was calculated using Equation ( 5), considering the PCE of the solar cells extracted from the J-V curves recorded under reverse and forward scan direction.Figure S10 (Supporting Information) depicts representative J-V curves (reverse and forward scan) of the most efficient PSCs fabricated by the usage of the DMF-based and GVLbased inks, while the corresponding electrical characteristics of the solar cells are tabulated in Table S4 (Supporting Information).
As it is perceived, both types of devices present mild hysteresis in their electrical characteristics, which is found on the order of 17% and 13% for the devices fabricated by the 1.8 m DMF-based and 0.8 m GVL-based inks, respectively.HI = ( PCE rev − PCE for ) ∕PCE rev (5)   After the evaluation of the performance of the PSCs in terms of extraction of their J-V curves, the quantum efficiency of the devices was also determined.Figure 5 presents both the incident photon-to-electron conversion efficiency (IPCE) and absorbed photon-to-current conversion efficiency (APCE) spectra of the solar cells, with the latter being calculated using Equation ( 6), revealing important information related to the nature of the charge injection efficiency of the devices.
Looking at both spectra, it can be perceived that the differences in the quantum efficiency of the solar cells are quite small, in agreement with the J-V characterization results.Both types of devices were able to attain IPCE values ranging from 80% to 90% in a great part of the visible spectrum (namely from 400 to 700 nm), while the APCE values were found to be on the order of almost 100% for the spectral region of 350 to 500 nm.Similar APCE values for PSCs employing CH 3 NH 3 PbI 3 as an active material have been recorded in previous investigations, underling the difficulty in further improvements of the performance of this type of solar cells in the case that a single absorber layer is employed. [37]erein, it is also noteworthy to mention that the calculated J SC values derived by integration of the IPCE of the solar cells are almost in total agreement with the corresponding values recorded in the J-V characterization of the devices.The slightly increased J SC attained by the PSCs developed using the DMF-based ink compared to the devices developed using the GVL-based ink was attributed to the slightly higher charge collection efficiency of the absorbed photons in the region of 550 to 800 nm.
Subsequently, electrochemical impedance spectroscopy (EIS) was applied to determine the charge transfer and recombination kinetics inside the PSCs.The Nyquist plots and Bode phase diagrams of the solar cells employing the perovskite absorbent layers developed by the 1.8 m DMF-based and 0.8 m GVL-based inks are presented in Figure 6, while the total ohmic series resistance (R S ), recombination resistance (R REC ) and average electron lifetime ( e ) obtained by the EIS analysis are tabulated in Table 4.
Looking at the Nyquist plots, it can be perceived that both spectra are characterized by the presence of two arcs, with one big arc to be located at the high-frequency domain and a much smaller   one at the low-frequency domain.According to previous investigations, the high-frequency arc is correlated with the bulk recombination inside the solar cell, while the low-frequency arc is developed due to both ionic and electronic charge accumulation at the interfaces of the perovskite absorbent layer with the chargeselective contacts. [38]Considering the aforementioned and by taking into account that the performance of these devices under their operation can be solely deduced from their impedance response at the high-frequency domain, [39] the simplified equiva-lent circuit of the type R S [CPE HF (R HF [CPE LF R LF ])] was applied to fit the experimental impedance data (see inset of Figure 6a).Under this assumption, the R S and R REC = R HF were extracted, giving evidence of the charge transport and recombination taking place inside the PSCs.Besides, as an extra indicator of the charge recombination rate taking place inside the PV devices,  e was calculated from the frequency value when the curve at the high-frequency regime of the Bode phase diagrams (Figure 6b) maximizes.By comparing the impedance data of the solar cells, it is demonstrated that the main differences appear in the values of R REC and  e , with the values of these parameters being higher in the case of the devices fabricated using the GVL-based ink.Namely, a decreased charge recombination rate inside these solar cells takes place, which is in line with the increased V OC that is demonstrated by these devices.The aforementioned observation is mainly attributed to the better morphological characteristics of the perovskite absorbent layer developed by the GVL-based ink compared to the case of DMF-based ink, as it is demonstrated by FE-SEM and optical microscopy analysis.This is because the uneven rod-shaped morphology developed by the usage of the DMF-based ink led to poor coverage of the underneath ETM, making easy the direct contact of the TiO 2 with the carbon materials, resulting in a low shunt resistance and the development of microshorts that disturb the operation of the solar cell as a photodiode.
Determining the degradation of the electrical characteristics of PSCs by their dark storage aging at ambient air conditions is considered the first crucial step in the log way of stability assessment of these devices.Herein, the solar cells developed by the usage of the 1.8 m DMF-based and 0.8 m GVL-based inks were aged according to the ISOS-D-1 protocol (storage of the devices in the dark, under T = 25 ± 5 °C and RH ambient), giving evidence on the tolerance of these devices to oxygen, moisture and other aggressive atmospheric components that are naturally presented in the air. [40]Figure 7 presents the variation of all the main PV parameters in time, namely J SC , V OC , FF and PCE, for both types of devices.For a better comparison, all the PV parameters were normalized to their initial values (i.e., for t = 0).
As it is evidenced by Figure 7, both types of unencapsulated solar cells demonstrate high stability under the conditions determined by the ISOS-D-1 protocol, retaining almost 95% of their initial light-to-electricity conversion efficiency after about 1000 h of their aging (T 95 = 1000 h).For both cases, the degradation of the performance of the devices in time was almost at the same rate, and it was attributed to the decrease of FF, while J SC and V OC did not present any notable variation.As it is demonstrated in previous studies, CH 3 NH 3 PbI 3 material is complexed with H 2 O molecules when exposed to ambient air, resulting in the development of hydrate products similar to (CH 3 NH 3 ) 4 PbI 6 •2H 2 O.This results in the deterioration of the perovskite crystal structure, leading to a decreased bulk photoconductivity and an increased charge recombination rate inside the solar cells, subsequently affecting the FF value (increase in the total series parasitic resistance and decrease in the shunt parasitic resistance). [41]o the final aim of the present investigation, the upscaling of the small-sized solar cells to the mini-module level (100 cm 2 aperture area) was evaluated using the 0.8 m GVL-based ink and a C-based HTM-free architecture as presented in Figure 8a.The mini-module was fabricated using a structure of 10 strips of cells connected in series, each of them having an active area of 5.24 cm 2 , giving a total module active area of 52.4 cm 2 (see Figure 8b).Figure 8c presents both reverse and forward I-V curves of the mini-module measured under real test conditions (900 W m −2 ), while the electrical characteristics of the same batched small-sized solar cell and mini-module are tabulated in Table 5.
As it is demonstrated, the mini-module attained a lightto-electricity conversion efficiency of 10.07% (stabilized, see Figure S11, Supporting Information), with a J SC, strip of 18.2 mA cm −2 , V OC, strip of 940 mV and FF of 0.53, recorded using reverse I-V scan.On the other hand, the device performance under forward I-V scan was found to be slightly reduced, namely, PCE of 9.19%, J SC , strip of 18.3 mA cm −2 , V OC, strip of 942 mV and FF of 0.48, i.e., giving a mild hysteresis on the device electrical characteristics, which was found on the order of 8.7%.To determine the upscaling losses, the scaling-up factor (f scaling-up ) was determined using Equation (7), where A is the active area of the cell/module.
By considering the performance of the small-sized PSCs and the corresponding of the perovskite mini-module, the f scaling-up was determined to be on the order of 8.3% rel dec −1 (performance loss due to the geometrical FF is excluded).Noteworthily, this is a quite satisfactory value compared to corresponding achievements usually reported in the literature, even when compared to the scaling-up of the conventional noble-metal HTM-based perovskite devices. [10,42]This achievement underlies that an effective upscaling of perovskite PVs can be attained by using the much lower toxicity and stable GVL-based ink and the ultralow-cost, simplified and industry-compatible C-based HTM-free architecture.

Conclusions
In summary, for the first time in inkjet-printing processing of PSCs, a low-molarity (0.8 m) nontoxic solvent based perovskite precursor ink is demonstrated to provide the fabrication of highly efficient and stable all-printed annealing-free perovskite PV devices under ambient atmosphere.GVL solvent was found as a quite promising green alternative to prepare lower toxicity and highly stable (stable for more than 2 months in storage) inkjetprintable inks that can enable the fabrication of high-quality antisolvent-free perovskite absorbent layers.Due to the combined solvent attributes of low vapor pressure and noncoordinating capability of the solvent molecules with the perovskite precursors, the development of high crystallinity, compact and with minimum coffee-ring defects perovskite absorbent layers (as also presented by in situ video microscopy) could be realized, without the need of applying an annealing procedure after the inkjetprinting process.The development of small-sized solar cells by the use of the proposed ink and an industry-compatible C-based HTM-free architecture demonstrated that devices with a high light-toelectricity conversion efficiency can be fabricated, attaining an even higher PCE than the performance of corresponding devices fabricated by the usage of a reference 2.25 times higher molarity perovskite precursor ink.Namely, the champion PCE of the small-sized solar cells surpassed the 13%, which is set among the highest reported values for the under-consideration PSC architecture employing an inkjet-printed perovskite absorbent layer.Besides, the PV devices presented high robustness under the conditions determined by the ISOS-D-1 protocol, where they could retain almost 95% of their initial performance after more than 1000 h of their aging.Finally, the ink was evaluated in the development of an all-printed perovskite mini-module of 100 cm 2 aperture area.The results showed that a 10%-efficient (stabilized) PV prototype can be fabricated, with the scaling-up losses to be determined on the order of 8.3% rel dec −1 per upscaled active area.The aforementioned findings are important since they provide a new insight into the worldwide effort given today toward the development of ultralow-cost, sustainable and scalable manufacturing of PSCs that would open the way for the widespread commercialization of this technology soon.
Fabrication of the Perovskite Solar Cells and Perovskite Mini-Module: The all-printed PVs were developed under ambient atmosphere (20-25 °C and 30-50% RH) using an inkjet-printer of the model Fujifilm Dimatix DMP-2850 employing a Dimatix Samba printhead, a homemade screen printer employing a 43T mesh screen and a dispense printer of the model FISNAR F4500N.
For the Small-Sized Perovskite Solar Cells: Initially, the FTO glass was mechanically scribed to attain a patterned electrode and subsequently cleaned using detergent, distilled water and isopropanol, followed by its calcination at 500 °C to remove any contaminants.Right after, the metal contacts of the under-development devices were fabricated using dispense-printing (speed of dispensing 5.5 mm s −1 and flow rate 0.03 mL min −1 ) and calcinating the system at 125 and 500 °C for 10 and 60 min, respectively.The hole-blocking layer (compact titanium dioxide, c-TiO 2 ) was developed on the top of the FTO glass by inkjet-printing (1335 dpi, singlepass) using an ink of titanium(di-isopropoxide) bis(2,4-pentanedionate) 75% in 2-propanol (Alfa Aesar) diluted (1/9 v/v) in a mixture of terpineol and acetonitrile (2:1 v/v) and calcinating the system at 500 °C for 10 min; the procedure of printing was applied five times to attain the appropriate c-TiO 2 film thickness.On the top of the c-TiO 2 layer, a mesoporous titanium dioxide (m-TiO 2 ) layer was fabricated by inkjet-printing (1955 dpi, quad-pass) a TiO 2 paste that was prepared using a simple chemical technique as reported in a previous work and calcinating the system at 500 °C for 30 min. [43]The formulation of the TiO 2 paste was modified compared to the referenced recipe to facilitate the paste inkjetprinting; ethanol was replaced by 2-propoxy-ethanol, while the obtained paste was further diluted (1:3 v:v) in 2-propoxy-ethanol.The active layer of the solar cells was developed using methylammonium lead triiodide (CH 3 NH 3 PbI 3 ) precursor inks composed of equal molar ratio methylammonium iodide (CH 3 NH 3 I) and lead iodide (PbI 2 ).To this aim, two different inks were prepared, one reference of 1.8 m CH 3 NH 3 PbI 3 in DMF and another of 0.8 m CH 3 NH 3 PbI 3 in GVL.The molarity of the inks in perovskite precursors was determined by their dissolution ability set by each applied solvent.Both inks were inkjet-printed (725 dpi, quad-pass) on m-TiO 2 after an initial optimization of the printing parameters evaluated in terms of PV performance; the DMF-based ink was printed at 60 °C and at 5 kHz frequency, while the GVL-based ink was printed at 50 °C and 2.5 kHz frequency.In both cases, the printing procedure was carried out on preheated substrates at 60 °C to eliminate the effect of moisture, according to previous studies. [44]The annealing of the perovskite absorbent layer was carried out at 100 °C for 20 min, while devices without annealing were also developed.On the top of the perovskite absorbent layer, a carbon layer was fabricated by screen-printing (single-pass, gap 3 mm) a carbon paste that demonstrates benign perovskite compatibility, prepared according to a previous report; [45] its calcination was carried out at 70 °C for 20 h.The aperture area of the small-sized solar cells was fixed at about 0.3 cm 2 .
For the Perovskite Mini-Module: The perovskite mini-module of 100 cm 2 (10 × 10 cm) aperture area was developed on the basis of the fabrication procedure described for the small-sized PSCs, using 10 strips of cells (active width set at 5.5 mm) connected in series (scribing step set at 10.5 mm), each strip having an active area of 5.24 cm 2 .Namely, the total active area of the mini-module was 52.4 cm 2 .
Characterization: An Ostwald viscometer of the model SCHOTT AVS310 was used to measure the viscosity of the perovskite precursor inks at printing temperature.A KSV Cam 100 contact angle optical measuring system equipped with a Hamilton 1001TPLT 1.0 mL threatened plunger syringe was used to measure the surface tension of the perovskite precursor inks and the wettability of the m-TiO 2 layer by the inks.More specifically, the pendant drop method was applied to determine the surface tension of the inks after Young-Laplace fitting (see Figure S12, Supporting Information), and the sessile drop method was applied for the wettabil-ity measurements. [46]XRD was used to determine the crystallinity of the perovskite absorbent layers, on a Bruker D8 Advance diffractometer with a diffracted beam monochromatic Cu Ka radiation source ( = 1.5496Å), from 5°to 60°, at a scan rate of 2°min −1 .FTIR spectroscopy was used to investigate the chemical interactions between the solvent molecules and perovskite precursors in the developed inkjet-printable inks, on a Jasco FT/IR-400 spectrometer, over the wavenumbers region between 2000 to 1500 cm −1 and at a resolution of 1 cm −1 . 1 H NMR spectra of the developed perovskite powders were recorded on a Bruker Advance III (Bruker BioSpin GmbH) DPX 600 spectrometer, using DMSO-d6 as a solvent.UV-Vis absorption spectra were collected to investigate the development of iodoplumbate complexes in the inkjet-printable inks diluted at 0.1 m, on a Hitachi U-1800 spectrophotometer, from 300 to 500 nm, using an interval wavelength of 5 nm.FE-SEM was applied to investigate the morphology of the perovskite absorbent layers, on a Zeiss SUPRA 35 VP microscope, while top-view optical images were also taken using the built-in camera of the inkjet printer.Diffuse reflectance spectroscopy was applied to determine the optical characteristics (absorbance and reflectance spectra) of the FTO glass/c-TiO 2 layer/m-TiO 2 layer/CH 3 NH 3 PbI 3 layer systems, as well as the sole perovskite absorbent layers, on a Jasco V-770 spectrophotometer equipped with an ISN-923 integrating sphere, from 300 to 800 nm, using an interval wavelength of 5 nm.For the small-sized PSCs, currentvoltage curves were recorded under simulated standard test conditions (AM1.5G, 1000 W m −2 , 25 ± 5 °C), using a solar simulator of the model Solar Light (16S-300), while the mini-module was characterized under real test conditions (45 ± 2°zenith angle, ≈900 W m −2 and 20 ± 5 °C).In both cases, the measurements were taken using a Keithley 2601 source meter at a scan rate of 250 mV s −1 (reverse and forward scans).The operational stability of the mini-module was determined under sunlight (≈900 W m −2 ) while holding the device near its maximum power point conditions (V = 6.2 V).The stability of the PV devices was assessed by dark storage aging under 25 ± 5 °C and RH ambient (ISOS-D-1), measuring their I-V characteristics for a total duration of about 1000 h, with time intervals being set at 168 h (every week).The quantum efficiency of the solar cells was determined using a ThetaMetrisis PM-QE system equipped with a Xenon light source and a filter monochromator of the model Oriel Cornerstone 260 1/4 m, which was controlled by the PM-Monitor software; the measurements were carried out from 300 to 800 nm, using a wavelength interval of 5 nm and a time delay of 200 ms.EIS was used to determine the charge transport and recombination kinetics inside the solar cells, on a Metrohm Autolab PGSTAT 128 N potentiostat/galvanostat; the measurements were performed in the dark, at room temperature, over the 1 MHz to 1 Hz frequency range, using −V OC forward bias and ±10 mV perturbation.

Figure 1 .Table 2 .
Figure 1.X-ray diffraction (XRD) patterns of the perovskite absorbent layer fabricated on the top of the m-TiO 2 layer using the 1.8 m dimethylformamide (DMF) based and 0.8 m gamma-valerolactone (GVL) based perovskite precursor inks, after 20 min annealing and without annealing, respectively.

Figure 2 .
Figure 2. a,b) Top-view and c,d) cross-section scanning electron microscopy (SEM) images of the perovskite layer fabricated on the top of the m-TiO 2 layer using the 1.8 m dimethylformamide (DMF) based and 0.8 m gamma-valerolactone (GVL) based perovskite precursor inks, respectively.

Figure 3 .
Figure 3. Light-harvesting efficiency (LHE) of the perovskite layer fabricated on the TiO 2 -based working electrode using the 1.8 m dimethylformamide (DMF) based and 0.8 m gamma-valerolactone (GVL) based perovskite precursor inks.

Figure 4 .
Figure 4. J-V curves (reverse scan) of the most efficient solar cells fabricated using the 1.8 m dimethylformamide (DMF) based (case of 20 min annealing of perovskite film) and 0.8 m gamma-valerolactone (GVL) based (case of annealing-free perovskite film) perovskite precursor inks.

Figure 6 .
Figure 6.a) Nyquist plots and b) Bode phase diagrams for the solar cells fabricated using the 1.8 m dimethylformamide (DMF) based and 0.8 m gamma-valerolactone (GVL) based perovskite precursor inks.

Table 4 .
Parameters obtained from the electrochemical impedance spectroscopy (EIS) analysis for the solar cells fabricated using the 1.8 m dimethylformamide (DMF) based and 0.8 m gamma-valerolactone (GVL) based perovskite precursor inks.Perovskite precursor ink R S [Ohm] R rec [Ohm]  e [μs]

Figure 7 .
Figure 7. Dark storage aging of the unencapsulated solar cells fabricated using a) 1.8 m dimethylformamide (DMF) based and b) 0.8 m gammavalerolactone (GVL) based perovskite precursor ink.

Figure 8 .Table 5 .
Figure 8. a) Schematic image showing the structure of the perovskite solar mini-module, b) image of the fabricated mini-module, and c) I-V curves of the mini-module characterized under real test conditions.

Table 1 .
Parameters related to the jettability of the perovskite precursor inks.

Table 3 .
Electrical characteristics of the solar cells fabricated using the 1.8 m dimethylformamide (DMF) based and 0.8 m gamma-valerolactone (GVL) based perovskite precursor inks.