Access to finely tuned thin films that can act as electron transport layer (ETL) and adapt to the absorber composition and whole cell fabrication process is key to achieve efficient perovskite-based solar cells. In this study, the growth of mixed niobium-titanium oxide (Nb-TiO2) thin films by atomic layer deposition and its use to extract photogenerated electrons is reported. Films were obtained at 200 °C from titanium (IV) i-propoxide, (t-butylimido)tris(diethylamido)niobium(V), and water by introducing Nb2O5 growth cycle in a TiO2 matrix. Process parameters (order of precursor introduction, cycle ratio) were optimized; the growth mechanism and the effective Nb incorporation were investigated by an in situ quartz crystal microbalance and x-ray photoelectron spectroscopy. The composition, morphology, structural, and optoelectronic properties of the as-deposited films were determined using a variety of characterization techniques. As a result, a fine control of the film properties (between TiO2 and Nb2O5 ones) could be achieved by tuning Nb content. To allow a successful implementation in solar devices, a comprehensive annealing study under several conditions (temperatures, various atmospheres) was conducted leading to an evolution of the optical properties due to a morphological change. Ultimately, the incorporation of these 15 nm-thick films in mesoscopic perovskite solar cells as ETL shows an improvement of the cell performances and of their stability with increasing Nb content, in comparison of both TiO2 and Nb2O5 pure compounds, reaching power conversion efficiency up to 18.3% and a stability above 80% of its nominal value after 138 h under illumination.

In the last decade, perovskite solar cell (PSC) has attracted scientists worldwide attention,1–3 both as single absorber solar cells and as top cell in tandem devices, reaching record values of 26.1% and of 33.9%, respectively.4 Today, upscale manufacturing and the long-term stability of PSCs remain key issues for widespread applications.5,6 The stability of the perovskite photovoltaic device is strongly affected by the interfaces between the charge transport layers [electron transport layer (ETL), hole-transport layer (HTL)] and the perovskite absorber.7 Among different charge transport layer materials, inorganic metal oxides (TiO2, SnO2, NiOx) have demonstrated better stability in comparison to the organic counterpart (PTAA, PCBM, Spiro-OMeTAD) due to higher resilience to heat, light, and the local chemical environment.8 A TiO2 compact layer combined with a mesoporous TiO2 (mp-TiO2) scaffold is the most efficient candidate so far for an ETL in mesoscopic PSCs.9 In order to optimize and adjust the ETL optoelectrical properties, doping TiO2 with certain foreign elements such as Y3+,10 Ta5+,11 or Nb5+ is a common approach to promote electron extraction. The strategy of incorporating niobium in TiO2 materials has been applied on ITO-free,12 mesoscopic13–16 and planar PSC,14,17–19 and leads to an improvement of power conversion efficiency (PCE),13–16,18–20 reduction of hysteresis,16,17 and stability improvement.13,17,18 A strong electronic coupling between the charge transport layer and the perovskite for efficient interfacial charge transfer is necessary, as well as a fine tuning of the Nb-incorporation and of the layer thickness for the optimization of the film properties. Both solution-based13,14,16,18,19 and physical vapor deposition12 methods have been used to prepare such Nb-TiO2 nanomaterials for PSC applications. In parallel, the ALD is a powerful technique for growing oxide layers in PSC due to its relatively low deposition temperature, uniform coverage, nano-level control of the film composition and thickness and its industrial scalability.8,21–28 ALD is an alternative to the chemical vapor deposition (CVD) method, where the substrate surface is alternately exposed to the vaporized precursor fluxes, separated by purging periods to eliminate gas-phase reactions and remove reaction byproducts. The self-limited adsorption of the precursor on the substrate surface provides indeed inherent control of the film thickness, excellent repeatability, and allow deposition of conformal and pinhole-free films with superior uniformity.29,30 Nb-incorporated TiO2 thin films can be accessed by ALD from various combinations of precursors, namely, Ti(OiPr)4 (TTIP)/niobium(v) ethoxide (NEO)/H2O,31 TiCl4/NEO/H2O,32 Ti(OMe)4/NEO/H2O,33 TiCl4/niobium(v) (t-butylimido)tris(diethylamido) (TBTDEN)/H2O,34 and Ti(NMe2)4/TBTDEN/H2O.35 While Nb-TiO2 thin films prepared by ALD have been applied in catalysis31 and thermoelectricity32 applications, there is no, to the best of our knowledge, report of their applications in PSCs.

In this work, we finely tune the incorporation of Nb in TiO2 thin films and probe its influence on the film properties and efficiency as ETL in PSCs. For this, an alternative combination of precursors, various sequence of precursor introduction, and cycle ratio is used. Their influence on thin film optical, structural, and electrochemical properties is determined, and the growth mechanism is investigated by an in situ Quartz crystal Microbalance (QCM). In the mp-PSC fabrication process, TiO2 is implemented as a bilayer, with a mesoscopic film grown on top of the Nb-TiO2 one, involving a high temperature step (500 °C). Thus, to ease the integration of such ALD layers in PSC but also other fabrication processes in general, the impact of various annealing conditions on the ALD-grown layers has been investigated. Finally, the compact ALD-grown layers have been successfully implemented as ETLs in complete mesoscopic PSCs. A clear positive impact of the Nb incorporation on the cell is observed with an efficiency increase of 2.1% (PCE up to 18.3%) and an improved stability (above 80% of its nominal value after 138 h under illumination).

Thin films were prepared in a BENEQ TFS-200 ALD reactor. (t-butylimido)tris(diethylamido)niobium(V) (((CH3CH2)2N)3Nb = NC(CH3)3, TBTDEN, min. 98%, STREM), titanium (IV) i-propoxide [Ti(OiPr)4, TTIP, min. 98%, STREM] and de-ionized water were used as Nb, Ti, and O sources, respectively. All chemicals were used without further purification. Nitrogen (N2, 99.9999%, Air Liquide) was used as both carrier and purging gas. TTIP and TBTDEN were heated in hot solid source systems Beneq HS300 at TTTIP = 85 °C and TTBTDEN = 70 °C, respectively, while de-ionized water was kept at room temperature. Experiments were performed at Tdep = 200 °C and the pressure in the reaction chamber was kept in the range of 1–2 mbar. Thin films were deposited simultaneously on 2 mm-thick sodalime glass, single-polished n-doped Si (100) wafer with native oxide top layer and fluorine-doped tin oxide (FTO) on 3 mm-thick sodalime glass.

Nb-TiO2 film depositions were performed with the following program: n × (n1.{TiO2} + {Nb2O5}) where n is the number of supercyle and n1 the number of TiO2 cycles in one supercycle. The n1 values of 1, 2, 5, 10, and 20 were investigated. Three precursor sequences were investigated and referred to as sequence A (TTIP pulse/N2 purge/H2O pulse/N2 purge/TBTDEN pulse/N2 purge/H2O pulse/N2 purge) ; sequence B (TTIP pulse/N2 purge/H2O pulse/N2 purge/H2O pulse/N2 purge/TBTDEN pulse/N2 purge); and sequence C (H2O pulse/N2 purge/TTIP pulse/N2 purge/TBTDEN pulse/N2 purge/H2O pulse/N2 purge. A titanium oxide growth cycle, i.e., {TiO2}, is described elsewhere.36 A niobium oxide growth cycle, i.e., {Nb2O5}, corresponds to {Nb2O5} = {[TBTDEN]/N2/H2O/N2 = [1.5/0.05/0.5/2]/5/1/5s}. “Combination” mode was chosen to ensure proper mass transport of TTIP and TBTDEN. “Combination” mode is composed of four steps ([t1/t2/t3/t4]). First, N2 is injected in the precursor hot source (t1), then all valves are kept closed (t2), the precursor is pulsed (t3). Finally, N2 is injected with simultaneous precursor pulse (t4).

Post-treatments were performed by annealing at various temperatures and atmospheres. Annealing treatments under ambient air were done in a Controller P 330 Nabertherm oven, at 300, 500, and 600 °C with a heating speed of 1.5 °C/s and a plateau of 20 min at the set temperature. Annealing treatments under N2 were done in a Jipelec Jetfirst oven, at 300, 500, and 600 °C with the same heating sequence. Annealing treatments under forming gas (FG, 95% N2–5% H2, Alphagaz 2 quality) were done in an AsOne Annealsys oven, at 400, 500, and 600 °C, with a heating ramp of few minutes and a plateau of 20 min.

Thin films were characterized on samples deposited on single-polished n-doped Si (100) wafers with native oxide layer on top (ellipsometry, GIXRD, XRR, XPS), sodalime glass (spectrophotometry) or FTO (cyclovoltammetry). Thin film thicknesses were determined by spectroscopic ellipsometry (SE) performed with a Horiba Jobin Yvon Uvisel 2 ellipsometer, and x-ray reflectivity (XRR) using a PANalytical Empyrean equipment using Cu-Kα radiations. The film thickness homogeneity was confirmed over the whole substrate area (15 × 15 cm2). SE measurements in the wavelength range of 200–1700 nm were modelled using Tauc-Lorentz oscillators functions. X-ray diffraction (XRD) studies were performed under grazing incidence (GIXRD) conditions for crystallinity determination and phase detection.

Thin film compositions were obtained using a XPS Thermo Scientific K-Alpha+ spectrometer equipped with a monochromated Al-Kα x-ray source for excitation at 1486.6 eV and using an x-ray spot size of 400 μm. Calibration of the spectrometer was done using Cu and Au samples following the ASTM-E-902-94 protocol. In complement to surface analyses, in-depth profiling was carried out using a monoatomic ion gun (Ar+) with an energy of 1000 eV, a 10 mA current intensity, and an ion gun orientation of 30° from the sample surface. The thermo scientific advantage© software and library were used for the XPS data treatment and peak fitting. C 1s peak positioned at 284.9 eV was taken as a reference to present the high energy resolution core level spectra of the other elements. The valence band structure has been investigated by UPS analyses conducted on a Thermo Scientific 250 Xi spectrometer with an He source providing HeI photons at 21.2 eV.

Transmittance and reflectance spectra were obtained using an Agilent Cary 500 UV-Vis-NIR spectrophotometer equipped with an Agilent Diffuse Reflectance accessory. Cyclovoltammetry measurements were carried out on an ALD-film/FTO/glass stack and performed with an EC-LAB SP150 Bio-logic potentiostat using 0.8–1 cm2 as an effective area. All films were too resistive to be characterized by a four-point probe or Hall effect measurements. A classical three-electrode setup has been used with a glass/FTO substrate covered by the ALD-film as a working electrode, a Pt wire as a counter electrode, and a saturated calomel electrode (SCE) as a reference electrode (E vs SHE = 0.241 V). Thin films were dipped in a Fe(CN)63− (0.1 M)/Fe(CN)64− (0.1M) electrolyte. For each sample, the current response was recorded during three potential cycles applied (range between +1.5 and −1.5 V, rate = 20 mV/s).

1. Fabrication

The NIP architecture perovskite solar cells (PSCs) were fabricated with the following stack: glass substrate, FTO, niobium-incorporated titanium dioxide (Nb-TiO2), mesoporous titanium dioxide (mp-TiO2), triple-cation perovskite absorber (Cs0,05(MA0,167FA0,833)0,95Pb(I0,842Br0,158)3, poly(triarylamine) (PTAA) HTL, and Au back electrode. Description of all fabrication steps and characterization data of the perovskite absorber are provided in (Figs. S13–S15 in the supplementary material).50 

2. Characterization

The current–voltage characterizations (JV) were performed with an Oriel solar simulator under 1000 W/m2 corresponding to AM1.5G spectrum at 25 °C on PSC that were first light-soaked at Voc for 20 s. The irradiance was calibrated using a Si cell with a KG5 filter, which spectral response is comparable to the one of PSCs. The scan is conducted in both directions, forward then reverse, with a rate of 20 mV/s and a voltage step of 20 mV. The min and max voltage values were set at −0.1 and +1.2 V, respectively, along with an auto detect Voc, i.e., stopping the measurement just after Voc. A rectangular mask with area of 0.09 cm2 was used during the measurement to control the illuminated area. The external quantum efficiency (EQE) was measured with an Oriel IQE2000 with a step of 20 nm. Indoor accelerated aging was conducted for 138 h under LED solar simulator which spectrum matches to the AM1.5G (including UV portion) in a climate chamber kept under N2 atmosphere at 20 °C. An MPPT is conducted all along the aging process and IV characterizations were recorded every 15 min following the protocol described above. This experimental setup corresponds to ISOS-L1 protocol defined in the consensus statement for PSCs reliability testing.37 

A control of the Nb-incorporation, i.e., its atomic amount and dilution, is required to finely tune the properties of the thin films. By ALD, this is usually achieved by a so-called supercycle strategy, which is the combination of growth cycles of binary materials, the right combination of precursors, and the order of introduction of the precursors that directly impacts the surface chemical reactions involved.

First, three precursor sequences (A, B, and C, described in Fig. 1) have been tested as these are known to impact the incorporation of the additional atom.36,38–40 Mass variations relying on molecule desorption and adsorption during the film deposition have been evidenced by recording in situ QCM measurements. The results obtained for the three sequences are presented in Fig. 1 and compared to the ones of pure TiO2 and Nb2O5 thin films.

FIG. 1.

Mass variations recorded by in situ QCM measurements for TiO2 (a), Nb2O5 (b) and the three precursor sequences A (c), B (d), and C (e). The pulse times are depicted in colored bands below the curves.

FIG. 1.

Mass variations recorded by in situ QCM measurements for TiO2 (a), Nb2O5 (b) and the three precursor sequences A (c), B (d), and C (e). The pulse times are depicted in colored bands below the curves.

Close modal

For those two latter, positive mass variations are recorded after each metallic precursor pulses (ΔmTiO21 = 0.85 a.u. after TTIP pulse, and ΔmNb2O51 = 0.82 a.u. after TBTDEN pulse), while the H2O pulse leads to a negative mass variation (ΔmTiO22 = −0.4 a.u., and ΔmNb2O52 = −0.25 a.u.). It results in a growth per cycle (GPC) of ΔmTiO20 = 0.45 a.u./cycle for TiO2 and ΔmNb2O50 = 0.57 a.u./cycle for Nb2O5. Mass variations that occur during the {Nb2O5} cycle introduction in TiO2 were recorded for n1 = 5. In the case of sequence A where a water pulse follows each TTIP or TBTDEN pulse, mass variations are similar to the ones observed for pure TiO2 and Nb2O5mATiO2 = 0.47 a.u./cycle and ΔmANb2O5 = 0.54 a.u./cycle). It shows that both TiO2 and Nb2O5 are growing indifferently on themselves or on the other. In other words, both TTIP and TBTDEN are reacting in a very similar way on Ti–OH and Nb–OH terminated surfaces. In the case of sequence B, the TBTDEN pulse is followed by a TTIP pulse. While the mass variation during TBTDEN pulse remains unchanged (ΔmBNb = 0.81 a.u.), mass variation during the TTIP pulse is significantly smaller than in sequence A (ΔmBTi = 0.35 a.u. instead of ΔmTiO21 = 0.85 a.u.). This means that TTIP is reacting either with the Nb-ligand terminated surface fragments, or with the remaining –OH groups, or both. Finally, sequence C involves a TTIP pulse followed by a TBTDEN pulse. The TBTDEN pulse leads to no change in the mass variation (ΔmCNb ∼ 0), which shows that TBTDEN almost does not react with the TTIP-saturated surface, and makes sequence C irrelevant for incorporation of Nb in TiO2.

In summary, different Nb insertion levels can be achieved by choosing the precursor sequences. Sequence C leads to little to no insertion of niobium atoms. Sequence B shows smaller growth per supercycle due to the poor reaction of TTIP on TBTDEN-saturated surfaces. For both sequences B and C, reaction mechanisms of the two successive metallic pulses cannot be fully elucidated by QCM measurements, while sequence A leads to well-described growth mechanisms similar to the ones of bare TiO2 and Nb2O5. These observations, and the relative low GPC of TiO2 and Nb2O5 that should prevent the formation of biphasic systems, suggest that the Nb incorporation can be more easily tuned with sequence A by adjusting the cycle ratio n1.

QCM measurements were recorded with sequence A for several n1 values (n1 = 1, 2, 5, 10, and 20). No influence is observed on neither the Nb2O5 cycle nor the growth of TiO2 cycles, which confirms the independence of the growth of the two oxides on each other (see in Fig. S1 in the supplementary material).50 Hence, sequence A was selected as the most appropriate one to finely incorporate Nb and will be used for the rest of this work.

1. Tuning of Nb incorporation level and influence on thin film properties

Using a supercycle strategy, the cycle ratio {TiO2}:{Nb2O5} modulates the atomic composition of the films.41 Films with sequence A and different cycle ratios (n1 = 1, 2, 5, 10, 20, i.e., {TiO2}:{Nb2O5} = (1, 2, 5, 10, 20):1) were prepared. The number of supercycle, n, was set to generate 15 nm-thick films, considering the target application of those films as ETL in PSC. Assuming a simple rule of mixture between TiO2 and Nb2O5, it corresponds to 450 cycles for TiO2, 350 cycles for Nb2O5, and {n, n1} = {200,1}, {140,2}, {72,5}, {40,10}, {20,20} for Nb-incorporated TiO2 thin films.

Final film thicknesses follow a simple rule of mixture, with intermediary GPC values for all Nb-TiO2 samples (see Fig. S2 in the supplementary material).50 It confirms the previous hypothesis, i.e., the independence of the growth of the two oxides on each other, which lead to a simple rule of mixture. XPS measurements were performed on two points of each sample and confirmed the expected homogeneity of the surface composition and the presence of all expected elements (Ti, Nb, O) along with superficial C (13–16 at.  %) and N (0.6–1.7 at.  %) contaminations (see Fig. S3 in the supplementary material,50 Table S1 in the supplementary material).50 High energy resolution core level spectra of Ti 2p, Nb 3d, O 1s, C 1s, and N 1s were recorded for all cycle ratios. Ti and Nb exhibit oxidation numbers of Ti (+IV) with binding energies (BEs) of 458.6 eV (2p3/2) and 464.5 eV (2p1/2); and Nb (+V) with BE of 207.3 eV (3d5/2) and 210.0 eV (3d3/2) (Fig. 2).35 The only significant modification observed with the cycle number variation is the relative intensity of Ti 2p and Nb 3d peaks, assessing the presence of a unique phase of Ti1−xNbxO2. The corresponding atomic percentages of Nb and Ti are then employed to calculate the doping fraction xNb, defined as x Nb = [ Nb ] [ Nb ] + [ Ti ]. Figure 2 presents the xNb fraction variation with respect to the dopant cycle ratio, defined as R = 1 / ( n 1 + 1 ). We can observe that xNb follows the rule of mixture (dashed line), confirming our hypothesis that both Nb2O5 and TiO2 grow the same way on top of each other than on themselves. As a consequence, we demonstrated that the Nb amount can be finely controlled by the cycle ratio in the ALD supercycle, from [Nb]/[Ti] = 0.05 (n1 = 20) to 0.76 (n1 = 1). A specific double structure is observed for the N contamination with two features (BE = 400.4 eV, 394.8 eV) attributed to remaining ligand fragments. The low energy feature decreases with the cycle ratio, suggesting it is due to the TBTDEN ligands while the high energy one remains constant. Despite the sputtering artefact leading to the reduction of both Nb and Ti in Nb(+IV) and Ti(+III), a constant composition is measured within the whole layer (see Fig. S4 in the supplementary material).50 Since the film thickness (15 nm) is within the range of the escape depth of photoelectrons (7–10 nm), almost the entire layer is probed at once before starting the profile. It is thus impossible to argue for a layered structure or a homogeneous one. The same conclusion rises from additional non-destructive profiling by angle-resolved measurements (from 20° to 60°) (see Fig. S5 in the supplementary material)50 where no drastic evolution of the composition is measured when probing either 2–3 nm or 7–10 nm. UPS measurements reveal an impact of the incorporation of Nb on the thin film electronic structure. Indeed, there is a shift of the valance band maximum (VBM) position and of the work function, which translates in a modification of the conduction band minimum (CBM) (see Fig. S6 in the supplementary material).50 

FIG. 2.

XPS measurements of the Ti 2p (a), Nb 3d (b), and N 1s (c) peaks of Nb-incorporated TiO2 samples for n1 = 0, 1, 2, 5, 10, 20. The evolution of the Nb doping fraction xNb calculated from the atomic percentage measured by XPS is depicted with respect to the dopant cycle ratio R (d).

FIG. 2.

XPS measurements of the Ti 2p (a), Nb 3d (b), and N 1s (c) peaks of Nb-incorporated TiO2 samples for n1 = 0, 1, 2, 5, 10, 20. The evolution of the Nb doping fraction xNb calculated from the atomic percentage measured by XPS is depicted with respect to the dopant cycle ratio R (d).

Close modal

Structurally, all thin films appear amorphous as determined by GIXRD. The cycle ratio seems to have the expected impact on the optical properties of the films, namely, the optical bandgaps of the ALD layers are evolving continuously between the bandgap of TiO2 Eg = 3.32 eV and the one of Nb2O5 Eg = 3.46 eV.42,43 Cyclovoltammetry studies show that all samples are n-type (presence of a reduction peak) and display a hole-blocking behavior (no oxidation peak),44 with very little impact of the Nb amount (see in Figs. S7–S10 in the supplementary material).50 

2. Impact of post-treatments on thin film properties

In the target application, the Nb-TiO2 thin films are covered with mp-TiO2, which fabrication requires a heating step at 500 °C under air. Hence, the influence of post-treatment on the film properties had to be investigated. More generally, as the application of those films is not limited to a single specific PSC architecture, the impact of various annealing post-treatments, i.e., at various temperatures and atmospheres, was investigated. In particular, the impact of annealing in ambient air at 300, 500, and 600 °C; in N2 at 300, 500, and 600 °C; and in forming gas (FG, H2 5% – N2 95%) at 400, 500, 600 °C, was explored.

A first impact of annealing is evidenced by GIXRD (see in Fig. S10 in the supplementary material).50 The TiO2 layer, amorphous as deposited at 200 °C, crystallizes at 300 °C in all atmosphere conditions tested in the anatase phase (ICDD 00-021-1272), with the specific main diffraction peaks (101), (004), (200). When Nb is incorporated, the crystallization leads to the same anatase phase as TiO2, for all Nb doping fraction investigated, from xNb = 0.04 (n1 = 20) up to xNb = 0.43 (n1 = 1). For n1 = 5, 10, 20, the Nb-TiO2 crystallizes at 300 °C in all atmospheres. Higher amounts of Nb atoms delay the crystallization that takes place at higher temperatures [at 400 °C (FG)- 500 °C (air and N2) for n1 = 2, at 600 °C in all atmospheres for n1 = 1]. A finer analysis shows a small shift towards lower angles of the anatase peaks for increasing Nb amount. It demonstrates an expansion of the unit cell, with a linear increase of its volume with respect to the Nb doping fraction.32 

Electrochemical characterization by cyclovoltammetry was made for the samples annealed at 500 °C. After annealing at 500 °C, all samples lose their hole-blocking behaviour with the emergence of an oxidation peak (∼1 mA after annealing in forming gas and ∼0.2 mA in air), while the reduction peaks remain (see in Fig. S11 in the supplementary material).50 This can be attributed to the crystallisation of the thin films, which is accompanied by its shrinkage and the apparition of cracks and grain boundaries, as seen by scanning electron microscopy (SEM) (see Fig. S12 in the supplementary material).50 SE and spectrophotometry measurements show a strong influence of the crystallization on the optical properties of the samples, especially on the band gap width as shown in Fig. 3. We observe an increase of 0.13 eV of the bandgap after the crystallization in the anatase phase. A smaller effect evidenced is the slight increase of the band gap with respect to the annealing temperature. At 500 °C, the temperature which would be reached during the integration in the solar cell, amorphous samples reach an average bandgap of Eg = 3.36 eV while the crystalline samples exhibit an average band gap of Eg = 3.50 eV. The continuous evolution of the bandgap between the value of TiO2 and the one of Nb2O5 with increasing Nb amount is kept for all annealing temperatures.

FIG. 3.

Evolution of the bandgap determined by SE with respect to the annealing temperature gathering the data obtained for all tested annealing atmospheres (air, N2, FG). Data are plot depending on the structural properties (amorphous, blue/black; crystalline, red/bright grey) with no regard of the Nb doping fraction or the annealing atmosphere conditions.

FIG. 3.

Evolution of the bandgap determined by SE with respect to the annealing temperature gathering the data obtained for all tested annealing atmospheres (air, N2, FG). Data are plot depending on the structural properties (amorphous, blue/black; crystalline, red/bright grey) with no regard of the Nb doping fraction or the annealing atmosphere conditions.

Close modal

Such differences could lead to significant effect on solar cell performance. Hence, the five Ti:Nb ratios (n1 = 1, 2, 5, 10, 20) along with the TiO2 and Nb2O5 references were then integrated as ETL in perovskite solar cells, and their performances were investigated in terms of efficiency and stability.

Perovskite-based solar cells incorporating ALD-grown Nb-TiO2 layers (n1 = 1, 2, 5, 10, and 20) as ETL were fabricated (see schematic representation of the full stack in Fig. 4), along with devices based on bare TiO2 as reference, but also bare Nb2O5 as this latter has been reported as efficient ETL for PSCs.45–47 For each condition, four cells were prepared and characterized by JV measurements in both directions forward (FW) and reverse (RV), external quantum efficiency, MPPT tracking during indoor accelerated ageing under continuous illumination to evaluate their relative power conversion efficiencies and stabilities.

FIG. 4.

J(V) (a) and solar cell performance characteristics [PCE, JSC, VOC and FF, (b)] of PSC with various ETL compositions. Schematic representation of the complete stack of PSC (c).

FIG. 4.

J(V) (a) and solar cell performance characteristics [PCE, JSC, VOC and FF, (b)] of PSC with various ETL compositions. Schematic representation of the complete stack of PSC (c).

Close modal

An important hysteresis between FW and RV measurements is observed for all samples with an average PCE difference of 3.1% (and a relative difference PC E RV PC E FW PC E RV = 18.2 %), which is common for fresh PSC (measurement at D0). In the following, we focus on RV measurements that show the best performances. First, when comparing at D0 PSC with bare TiO2 or bare Nb2O5 (respectively blue and red in Fig. 4), only a slight increase of performance is observed: average PCE is 16.2% for TiO2 and 16.45% for Nb2O5, with similar JSC and VOC values, which is accordance with the literature.48 For all PSCs with Nb-TiO2 layers, a continuous increase of PCE is observed from the TiO2 value up to n1 = 1 with an average PCE of 17.96% and a champion cell at 18.30%. The open-circuit voltage VOC shows an increase with the Nb amount, recovering the value of both bare TiO2 and Nb2O5 for n1 = 1 and 2 around 1.10 V. The short-circuit current JSC also exhibits an increase from 20.88 mA/cm2 for TiO2 to 21.98 mA/cm2 for n1 = 1. The fill factor (FF) shows no clear trend, although the best values are obtained for Nb-TiO2 cells (n1 = 10, 5, and 1) with FF values above 75%.

To investigate the impact of those newly developed ETL layers on the PSC stability, two different methods were employed to determine the evolution with time of the solar cell performances.

The first common method to study ageing of PSC is referred as shelf ageing: PSC are stored in the dark and under vacuum for several days and J(V) characteristics are measured. In our case, PSC performances were determined at day 0, day 10, day 17, and day 61. Figure 5 shows the results between at day 0 and day 61 for TiO2, n1 = 10, 5, and 1. Hysteresis for PCE between FW and RV measurements after 61 days disappears. PCE decays with time for all samples, although the incorporation of niobium appears to stabilize the cell. Indeed, average PCE of TiO2 drops to 14.96% while the one of n1 = 1 remains at 17.55%, with a champion cell at 18.2%. For the PSC with the composition n1 = 1, there is no hysteresis after 61 days. Moreover, the J(V) curves after 61 days are very similar to the RV curve at day 0, showing a good stability over time.

FIG. 5.

Evolution of PCE for both FW and RV measurements at day 0 and day 61 of PSC with bare TiO2 and Nb-TiO2 with n1 = 10, 5, and 1 (a). FW and RV J(V) curves for n1 = 1 measured at day 0 and day 61 (b).

FIG. 5.

Evolution of PCE for both FW and RV measurements at day 0 and day 61 of PSC with bare TiO2 and Nb-TiO2 with n1 = 10, 5, and 1 (a). FW and RV J(V) curves for n1 = 1 measured at day 0 and day 61 (b).

Close modal

The second aging method that corresponds to ISOS-L1 protocol defined in the consensus statement for PSCs reliability testing,37 is under continuous illumination for 138 h, maximum power point tracking (MPPT) and JV measurements taken every 15 min. This protocol was applied on PSC with pure TiO2 for reference, n1 values of 20 and 2, and pure Nb2O5. EQE were measured both before (beginning of the life, BOL) and after (end of life, EOL) the aging (Fig. 6).

FIG. 6.

Evolution of normalized PV parameters [JSC (a), VOC (b), FF (c), and PCE (d)] with time under constant illumination and MPPT of PSC with bare TiO2, bare Nb2O5 and Nb-TiO2 with n1 = 2, 20. Measurements were done every 15 min; EQE measurements at BOL and EOL (e).

FIG. 6.

Evolution of normalized PV parameters [JSC (a), VOC (b), FF (c), and PCE (d)] with time under constant illumination and MPPT of PSC with bare TiO2, bare Nb2O5 and Nb-TiO2 with n1 = 2, 20. Measurements were done every 15 min; EQE measurements at BOL and EOL (e).

Close modal

Device based on pure TiO2 exhibit a quick loss of performance reaching 80% of its nominal PCE value within 10 h and showing no sign of stabilization, reaching 55% after 138 h [see Fig. 6(d)]. All the cell parameters (JSC, VOC, and FF) present similar decreasing trends, reaching JSC = 81%, VOC = 87% and FF = 78% of their nominal values after 138 h. Pure Nb2O5-based PSC efficiency shows an even faster deterioration within short aging period, reaching 80% after 3.5 h and 64% after 16 h. However, after 16 h the performances increase and stabilize above 70% from 62 h up to 138 h under illumination, assessing a better stability than TiO2 references. The PSC with Nb-TiO2 layers show an intermediary behaviour. For n1 = 20 (low Nb content), deterioration seems to be slower than for pure TiO2 for the first 15 h. However, it exhibits a continuous decrease, to finally reach similar performances than TiO2 after 138 h (58%). Finally, for n1 = 2 (high Nb content) PCE is stabilised after 8 h at 77% and slightly recovers to remain stable around 80% of its nominal value up to 138 h. This evidences a better stability of those cells compared to Nb2O5 ones and suggests the existence of an optimal Nb content for PSC stability. After the initial strong decrease, which is commonly attributed to interfaces,49 the degradation rate can evaluated by a linear fit between 20 and 70 h. In that case, PSC with pure TiO2 and n1 = 20 show a strong degradation with negative rates (−0.2 and −0.32%/h, respectively) while PSC with n1 = 2 and Nb2O5 exhibit positive rates, i.e., recovery (+0.01 and +0.13%/h, respectively). The evolution of JSC shows a huge drop for PSC with pure TiO2 (80% of its nominal value after 138 h) while other samples remain above 90% with a record value of 96% for n1 = 2. VOC and FF also show a recovery followed by a stabilization after an initial decrease for films with high Nb contents (n1 = 2 and Nb2O5) and a continuous degradation for low-Nb content (TiO2, n1 = 20). Improved stability with Nb incorporation is confirmed by EQE measurements before (BOL) and after (EOL) 138 h of illumination. PSC with pure TiO2 or Nb2O5 layers suffer loss in EQE around 20% and 13%, respectively, while PSC with Nb-TiO2 (n1 = 2) loses only 8% and remains near 80%.

In summary, ALD at 200 °C has been successfully used to grow Nb-TiO2 thin films with a very fine tuning of the film properties. In situ QCM measurements show that both TiO2 and Nb2O5 have stable GPC values (0.47 and 0.54 a.u./cycle, respectively) when grown either on top of themselves or on top of one another. It allows a fine control of the film composition and thickness, which are critical for tuning film properties. Film thickness and incorporation level of Nb are confirmed by ex situ ellipsometry, XRR, and XPS measurements. The homogeneity both in-depth and on the surface of the samples is insured as confirmed by multiple points, high resolution and angle-resolved XPS, along with the presence of a single amorphous phase with very low contamination.

All films remain too resistive for Hall effect measurements, but the Nb incorporation modifies the film electronic band structure as shown by UPS measurements. Films display a hole-blocking behavior and appear appropriate for an application as ETL in mesoscopic PSC. To mimic the PSC fabrication process, films were annealed at 500 °C. It leads to crystallization in the anatase phase, an increase of the bandgap of 0.13 eV and a minimal loss of the hole-blocking behavior.

When integrated as ETL in mesoscopic PSCs, the Nb-TiO2 films demonstrate a continuous enhancement of the cell performances with respect to the Nb amount, up to +7% of the TiO2 value for n1 = 1, reaching a PCE of 18.3%. More importantly, the stability of the PSCs is largely improved by the incorporation of Nb. Indeed, 80% of PSC PCE nominal value is conserved after 138 h under illumination for cell with Nb-TiO2 layer (n1 = 2), to be compared with 55% for TiO2 reference.

The improvement of both performance and stability of PSCs by the incorporation of Nb in a TiO2 layer grown by ALD in soft conditions represents a very promising result, considering that ALD can easily be implemented in large area systems.

The authors would like to thank J. Cardin, C. Labbe, and G. D. Gesesse (CIMAP, UMR6252, Caen, France) for their assistance in fitting SE data. This work was supported by the French Government in the frame of the program of investment for the future (Programme d’Investissement d’Avenir ANR-IEED-002-01) and the French Agence Nationale de la Recherche under Contract Nos. HANAMI ANR-17-CE09-0022.

The authors have no conflicts to disclose.

Thomas Vincent: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (lead); Writing – review & editing (equal). Damien Coutancier: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Pia Dally: Data curation (equal); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (equal); Visualization (lead); Writing – review & editing (equal). Mirella Al Katrib: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (supporting); Validation (equal); Visualization (lead); Writing – review & editing (equal). Mathieu Frégnaux: Data curation (supporting); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (supporting); Visualization (supporting); Writing – review & editing (equal). Stefania Cacovich: Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (supporting); Writing – review & editing (equal). Frédérique Donsanti: Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (supporting); Writing – review & editing (equal). Armelle Yaïche: Data curation (equal); Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (equal); Writing – review & editing (equal). Karim Medjoubi: Data curation (equal); Formal analysis (equal); Investigation (supporting); Methodology (equal); Validation (equal); Visualization (equal); Writing – review & editing (equal). Thomas Guillemot: Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (supporting); Writing – review & editing (equal). Marion Provost: Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Validation (supporting); Writing – review & editing (equal). Jean Rousset: Formal analysis (supporting); Investigation (supporting); Methodology (supporting); Project administration (equal); Resources (equal); Validation (supporting); Writing – review & editing (equal). Muriel Bouttemy: Formal analysis (supporting); Investigation (supporting); Methodology (equal); Project administration (equal); Resources (equal); Validation (supporting); Writing – review & editing (equal). Nathanaelle Schneider: Conceptualization (lead); Data curation (supporting); Formal analysis (equal); Funding acquisition (lead); Investigation (lead); Methodology (lead); Project administration (lead); Resources (lead); Validation (lead); Writing – original draft (supporting); Writing – review & editing (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
L.
Etgar
,
P.
Gao
,
Z.
Xue
,
Q.
Peng
,
A. K.
Chandiran
,
B.
Liu
,
Md.K.
Nazeeruddin
, and
M.
Grätzel
,
J. Am. Chem. Soc.
134
,
17396
(
2012
).
2.
J.
Burschka
,
N.
Pellet
,
S.-J.
Moon
,
R.
Humphry-Baker
,
P.
Gao
,
M. K.
Nazeeruddin
, and
M.
Grätzel
,
Nature
499
,
316
(
2013
).
3.
C.
Dong
,
Z.-K.
Wang
, and
L.-S.
Liao
,
Energy Technol.
8
,
1900804
(
2020
).
4.
“Best Research-Cell Efficiency Chart,” (n.d.).
5.
S. S.
Dipta
and
A.
Uddin
,
Energy Technol.
9
,
2100560
(
2021
).
6.
P.
Zhang
,
M.
Li
, and
W.-C.
Chen
,
Front. Chem.
10
, 802890 (
2022
).
7.
P.
Schulz
,
D.
Cahen
, and
A.
Kahn
,
Chem. Rev.
119
,
3349
(
2019
).
8.
S. S.
Shin
,
S. J.
Lee
, and
S. I.
Seok
,
Adv. Funct. Mater.
29
,
1900455
(
2019
).
9.
P. S. C.
Schulze
et al,
ACS Appl. Mater. Interfaces
9
,
30567
(
2017
).
11.
X.
Feng
,
K.
Shankar
,
M.
Paulose
, and
C. A.
Grimes
,
Angew. Chem. Int. Ed. Engl.
48
,
8095
(
2009
).
12.
S.
Goetz
et al,
ACS Appl. Energy Mater.
5
,
5318
(
2022
).
13.
Y.
Numata
,
R.
Ishikawa
,
Y.
Sanehira
,
A.
Kogo
,
H.
Shirai
, and
T.
Miyasaka
,
J. Mater. Chem. A
6
,
9583
(
2018
).
14.
G.
Xiao
,
C.
Shi
,
K.
Lv
,
C.
Ying
, and
Y.
Wang
,
ACS Appl. Energy Mater.
1
,
2576
(
2018
).
15.
Y.
Sanehira
,
N.
Shibayama
,
Y.
Numata
,
M.
Ikegami
, and
T.
Miyasaka
,
ACS Appl. Mater. Interfaces
12
,
15175
(
2020
).
16.
M.
Yu
,
H.
Sun
,
X.
Huang
,
Y.
Yan
, and
W.
Zhang
,
Nanoscale Res. Lett.
15
,
135
(
2020
).
17.
Y.
Lv
,
B.
Cai
,
Q.
Ma
,
Z.
Wang
,
J.
Liu
, and
W.-H.
Zhang
,
RSC Adv.
8
,
20982
(
2018
).
18.
G.
Yin
,
J.
Ma
,
H.
Jiang
,
J.
Li
,
D.
Yang
,
F.
Gao
,
J.
Zeng
,
Z.
Liu
, and
S. F.
Liu
,
ACS Appl. Mater. Interfaces
9
,
10752
(
2017
).
19.
B.-X.
Chen
,
H.-S.
Rao
,
W.-G.
Li
,
Y.-F.
Xu
,
H.-Y.
Chen
,
D.-B.
Kuang
, and
C.-Y.
Su
,
J. Mater. Chem. A
4
,
5647
(
2016
).
20.
M.
Abulikemu
,
M. L.
Tietze
,
S.
Waiprasoet
,
P.
Pattanasattayavong
,
B. E. A.
Tabrizi
,
V.
D’Elia
,
S. Del
Gobbo
, and
G. E.
Jabbour
,
ACS Omega
7
,
6616
(
2022
).
22.
K.
Deng
and
L.
Li
,
Adv. Mater. Interfaces
3
,
1600505
(
2016
).
23.
Z.
Xing
,
J.
Xiao
,
T.
Hu
,
X.
Meng
,
D.
Li
,
X.
Hu
, and
Y.
Chen
,
Small Methods
4
,
2000588
(
2020
).
24.
V.
Zardetto
,
B. L.
Williams
,
A.
Perrotta
,
F. Di
Giacomo
,
M. A.
Verheijen
,
R.
Andriessen
,
W. M. M.
Kessels
, and
M.
Creatore
,
Sustain. Energy Fuels
1
,
30
(
2017
).
25.
J. A.
Raiford
,
S. T.
Oyakhire
, and
S. F.
Bent
,
Energy Environ. Sci
13
,
1997
(
2020
).
26.
W.
Zhao
and
Y.
Duan
,
Adv. Photonics Res.
2
,
2100011
(
2021
).
27.
Y.
Yang
et al,
J. Mater. Chem. C
10
,
819
(
2022
).
28.
Y.
Zhang
,
Y.
Yang
,
M. T.
Mbumba
,
M. W.
Akram
,
E. K.
Rop
,
L.
Bai
, and
M.
Guli
,
Solar RRL
6
,
2200823
(
2022
).
29.
R. W.
Johnson
,
A.
Hultqvist
, and
S. F.
Bent
,
Mater. Today
17
,
236
(
2014
).
30.
Y.
Zhao
et al,
Chem. Soc. Rev.
50
,
3889
(
2021
).
31.
A. G.
Hufnagel
,
S.
Häringer
,
M.
Beetz
,
B.
Böller
,
D.
Fattakhova-Rohlfing
, and
T.
Bein
,
Nanoscale
11
,
14285
(
2019
).
32.
J.-P.
Niemelä
,
H.
Yamauchi
, and
M.
Karppinen
,
Thin Solid Films
551
,
19
(
2014
).
33.
V.
Pore
,
M.
Ritala
,
M.
Leskelä
,
T.
Saukkonen
, and
M.
Järn
,
Cryst. Growth Des.
9
,
2974
(
2009
).
34.
G.
Luka
,
L.
Wachnicki
,
R.
Jakiela
, and
E.
Lusakowska
,
J. Phys. D: Appl. Phys.
48
,
495305
(
2015
).
35.
W. J. H.
(Willem-J) Berghuis
,
J.
Melskens
,
B.
Macco
,
S. B.
Basuvalingam
,
M. A.
Verheijen
, and
W. M. M.
(Erwin) Kessels
,
J. Vac. Sci. Technol. A
38
,
022408
(
2020
).
36.
D.
Coutancier
,
S.-T.
Zhang
,
S.
Bernardini
,
O.
Fournier
,
T.
Mathieu-Pennober
,
F.
Donsanti
,
M.
Tchernycheva
,
M.
Foldyna
, and
N.
Schneider
,
ACS Appl. Mater. Interfaces
12
,
21036
(
2020
).
37.
38.
J. W.
Elam
and
S. M.
George
,
Chem. Mater.
15
,
1020
(
2003
).
39.
H.
Le Tulzo
,
N.
Schneider
,
D.
Lincot
,
G.
Patriarche
, and
F.
Donsanti
,
J. Vac. Sci. Technol. A
36
,
041502
(
2018
).
40.
J.
Kruszyńska
et al,
Adv. Mater. Interfaces
9
,
2200575
(
2022
).
41.
D. Y.
Lee
,
J.-H.
Park
,
Y.-H.
Kim
,
M.-H.
Lee
, and
N.-I.
Cho
,
Curr. Appl. Phys.
14
,
421
(
2014
).
42.
R.
Khan
,
H.
Ali-Löytty
,
J.
Saari
,
M.
Valden
,
A.
Tukiainen
,
K.
Lahtonen
, and
N. V.
Tkachenko
,
Nanomaterials
10
,
1567
(
2020
).
43.
S. B.
Basuvalingam
,
B.
Macco
,
H. C. M.
Knoops
,
J.
Melskens
,
W. M. M.
(Erwin) Kessels
, and
A. A.
Bol
,
J. Vac. Sci. Technol. A
36
,
041503
(
2018
).
44.
G. T.
Tractz
,
S. R.
Masetto Antunes
,
G. A.
Rodrigues Maia
,
H.
de Santana
,
M.
de Fátima Oliveira
, and
P. R. P.
Rodrigues
,
Opt. Mater.
121
,
111513
(
2021
).
45.
Y.
Guo
,
J.
Tao
,
J.
Jiang
,
J.
Zhang
,
J.
Yang
,
S.
Chen
, and
J.
Chu
,
Sol. Energy Mater. Sol. Cells
188
,
66
(
2018
).
46.
X.
Liu
,
Y.
Xiao
,
Q.
Zeng
,
J.
Jiang
, and
Y.
Li
,
J. Phys. Chem. Lett.
10
,
6382
(
2019
).
47.
R. D.
Chavan
,
N.
Parikh
,
M. M.
Tavakoli
,
D.
Prochowicz
,
A.
Kalam
,
P.
Yadav
,
P. H.
Bhoite
, and
C. K.
Hong
,
Adv. Mater. Interfaces
8
,
2100177
(
2021
).
48.
S. L.
Fernandes
,
A. C.
Véron
,
N. F. A.
Neto
,
F. A.
Nüesch
,
J. H.
Dias da Silva
,
M. A.
Zaghete
,
C. F.
de
, and
O.
Graeff
,
Mater. Lett.
181
,
103
(
2016
).
49.
J. A.
Christians
,
P.
Schulz
,
J. S.
Tinkham
,
T. H.
Schloemer
,
S. P.
Harvey
,
B. J.
Tremolet de Villers
,
A.
Sellinger
,
J. J.
Berry
, and
J. M.
Luther
,
Nat. Energy
3
,
68
(
2018
).
50.
See supplementary material online for additional data on substrate preparation, in situ QCM measurements (experimental details, mass variations during Nb-TiO2 film growth (precursor sequence A, various n1 values), characterizations of as-deposited (GPC measured by XRR and SE, surface homogeneity assessment by XPS, XPS depth profiling, angle-resolved XPS measurements, transmission-reflection measured by spectrophotometry, light absorption spectra, SE data and evolution of the optical bandgap, cyclovoltammetry), and annealed (GIXRD diffractograms, cyclovoltammetry, SEM) thin films, and solar cells (description of all fabrication steps, structural characterization of the perovskite absorber by SEM, XRD, UV-Vis absorption spectra of the stacks).

Supplementary Material