Low dimensional perovskites belong to a large family of structures which read as R₂(A)_n−1M_nX_3n+1, where R is the bulky organic molecule, A the small organic cation, M the divalent metal, X the halogen and n the number of inorganic layers (Mitzi, 2001). For n = 1, one layer of corner-shared MX64- octahedral is sandwiched in between two arrays of large organic spacer. Figure 1A shows the molecular structure of the organic ammonium salts (Lc and Lf) investigated in this work. They both feature a nine-carbon alkyl chain, with the partial substitution of hydrogen for fluorine in Lf cation. The UV-Vis absorbance spectra at r.t of the corresponding (Lf)₂PbI₄ and (Lc)₂PbI₄ 2D perovskites in thin film are shown in Figure 1B. The strong excitonic peak dominates the spectra in both cases, a characteristic feature of 2D structures. Interestingly, despite the fact that the organic component does not contribute to the frontier orbitals of the 2D perovskite (Traore et al., 2018), we found a stark blue-shift of the excitonic resonance as a result of the presence of fluorine in the organic spacer. The lowest exciton band of Lc and Lf appear at 2.4 eV (513 nm) and 2.6 eV (483 nm), respectively, with a blue-shift of 0.2 eV. Further comparison with the well-studied 2D perovskite (BuA)₂PbI₄ (based on n-butylammonium cation, BuA) shows exactly the same exitonic peak position as (Lc)₂PbI₄ (see Figure S1). In parallel, also the continuum feature related to the onset of the band-to-band transitions, is blue shifted by compatible quantity (above 2.7 eV and 2.9 eV for the Lc and Lf, respectively). We note that this blue-shift is also present in photothermal deflection spectroscopy measurements (Figure 1C), which also allow us to quantify the energetic disorder in the samples by determining the Urbach Energy (E_U). It is interesting that despite the similarities in the 2D structure of the films, Lf exhibits a larger E_U of 29.85 meV, as compared to only 26.15 eV for Lc. This suggests that the energetic disorder in Lf is higher, possibly due to larger distortion of the perovskite structure (see Figure 1D). It is also worth mentioning that form a microscopic point of view, different Urbach Energy can also partially derive from slight differences in film morphology and crystal quality of the two samples which are averaged out in the absorption measurements. Figure 1E shows the X-Ray Diffraction (XRD) patterns of the two films. The low angles diffraction signals are indicative of 2D perovskite n = 1 structures (Cao et al., 2015; Mao et al., 2016; Stoumpos et al., 2016). Specifically, the patterns are dominated by equally spaced (00l) reflections, peaking at 4.5° and 3.4° for (Lc)₂PbI₄, and (Lf)₂PbI₄, respectively. The reflections of the other (hkl) planes are strongly suppressed, confirming the remarkably preferred orientation of the inorganic planes, aligning parallel to the substrate (Mao et al., 2016; Stoumpos et al., 2016). The distance between the inorganic planes results in spacing of 26.75 and 20.07 Å for (Lf)₂PbI₄ and (Lc)₂PbI₄, respectively. The XRD pattern of (BuA)₂PbI₄ film is used for comparison (Figure S2).

These results raise an important question: what is the origin of the shift in the absorbance spectra, which induces a 0.2 eV blue-shift of the excitonic resonance. Conceptually, such change in the optical properties can be explained either as an opening of the electronic band gap upon introduction of Lf, or as a decrease of the exciton binding energy, or a combination of the two effects (see Figure 1D). The determination of the band gap from the UV-vis spectrum can be challenging in the case of excitonic materials, as the ones investigated here. In fact, the exact rise of the band-to-band transition is hidden by the excitonic resonance. Tentative estimation of the band gap via Tauc plot is reported in Figure S3, which suggests a blue-shift of 0.1 eV going from Lc to Lf. To support this analysis, we performed periodic Density Functional Theory (DFT) calculations, which directly evaluate the single particle band gap, in absence of excitonic effects. In the present case, the calculation is complicated by the lack of structural information for the Lf compound. In fact, while structural models based on XRD measurements are fully available from the literature for the Lc compound (Billing and Lemmerer, 2008), no structural models we available for the Lf one, with the only structural information available represented by the reported increase in interlayer distance illustrated in Figure 1E. We hence started from the orthorhombic structure reported for the Lc compound (Billing and Lemmerer, 2008) and we provided reference model for the Lf compound, by substituting the hydrogen with fluorine in the cation and increasing the interlayer distance, consistently with the datum in Figure 1E. Then, for both Lc and Lf, we performed structural relaxation of all the atomic parameters, keeping the lattice and space group fixed, so to obtain the equilibrium structure of the two compounds at zero K temperature. To accurately describe van der Waals interactions, which are expected to play a role on intermolecular interactions between organic chains, we used VDW-DF2 functional, which has been explicitly designed for this purpose (Lee et al., 2010). We then computed the single particle band gaps using the PBE exchange-correlation functional, including Spin-Orbit Coupling, which is known to largely affect the electronic properties of lead-based HP (Even et al., 2013), but without affecting the optical anisotropy of 2D perovskites (Even et al., 2012). For Lc and Lf, we found band gap, respectively, of 1.34 and 1.46 eV. The computed single particle band gaps are severely underestimated with respect to the experimental data, because of lack of accurate treatment of electronic exchange and correlation (Quarti et al., 2018), but on the other hand, it provides a blue-shift of the band gap of 0.12 eV going from Lc to Lf, which goes in the same direction of the Tauc plot analysis. In turn, this band gap opening upon cation substitution is very well-rationalized in terms of increased structural distortions in the inorganic plane, in agreement with the finding of Knutson et al. (2005). The tilting in the octahedra is conveniently described in terms of the angles β and δ as showed in the inset of Figure 1F. In the figure, the band gap of a purely structural model of 2D perovskite is mapped as function of the values of these two angles at the same level of theory (Pedesseau et al., 2016). As shown, the more the structure distorts (β and δ deviating from 180° and 0°, respectively), the more the band gap increases, because of the reduced atomic wavefunction overlap, and consequent decrease of band valence and conduction bandwidth (Umebayashi et al., 2003; Quarti et al., 2018; Traore et al., 2018). Nicely, the band gap computed for the structure of Lc and Lf well-fit in the 2D map in Figure 1F. As final remark, the lack of structural information for Lf and the use of in-plane lattice parameters from Lc can clearly influence the results of the present analysis. To check this possibility, we further relaxed both the atomic positions and the lattice parameters. In the present case, the obtained band gaps, 1.29 and 1.43 eV for Lc and Lf, respectively, are slightly different than the results obtained by fixing the lattice, but the blue-shift from the former to the latter compound is confirmed, hence reinforcing the present analysis. This structural effect is hence consistent with the band gap opening mechanism reported in Figure 1D. However, a possible effect of the organic cation on the exciton binding energy remains to be checked. Quantitative estimation of the exciton binding energy is on the other way a very hard task, as the inclusion of the electron-hole interaction requires the use of complex approaches, based on the Time-Dependent (TD)DFT method (Quarti et al., 2018) or on the solution of BSE performed on converged single particle electronic structure, including spin-orbit coupling (Giorgi et al., 2018). On the other hand, the computational cost associated to these calculations is very large and they are not yet able to describe the well-known deviations from 2D Rydberg series, making these methods unaffordable for the compounds investigated here and the requested level of accuracy. Alternatively, one can address changes in exciton binding energy by estimating the change in the dielectric inhomogeneity along the stacking axis, which enhance the screening between the photo-excited electron-hole pair by the remaining electrons (Katan et al., 2019). It is in fact reasonable to expect some change in the dielectric response, as results of the introduction of electron rich fluorine atoms. To this aim, we adopted the formalism developed in Even et al. (2014) and Sapori et al. (2016) which consists in performing DFT calculations in presence of external fields and to determine the profile of the dielectric response from the electron density re-organization. More importantly, by determining the dielectric profile along the normal to the inorganic layer, one is able to distinguish between the dielectric response of the inorganic and of the organic components. The dielectric profile computed for both Lc and Lf following this approach is shown in Figure 1G, and indicates that only very limited changes in the dielectric response are found upon introduction of fluorine atoms. In comparison, changes of the dielectric constant of a factor of 3 where reported upon intercalation of molecular I₂ into 2D perovskites, with corresponding sizable change in the exciton binding energy (Smith et al., 2017). We conclude that the blue-shift observed in Figure 1B can be mainly attributed to a band gap opening, with negligible effects on the exciton binding energy.

Interesting differences can be observed in the optical response as a function of temperature, following the partial fluorination of the organic spacer. UV-Vis and PL measurements were carried out going from room temperature down to 80 K (see experimental methods for details). The results for (Lc)₂PbI₄ and (Lf)₂PbI₄ thin films are collected in Figure 2. The larger population of phonons found with increasing the temperature results in more frequent electron-phonon scattering events, which in turn determines the larger bandwidth of the peak observed at r.t., compared to low temperature measurements. Remarkably, in the case of fully-hydrogenated cation-based material, (Lc)₂PbI₄, a second peak at low wavelength (483 nm) appears between 240 and 200 K along with a decrease in the intensity of the main peak at 2.4 eV (517 nm) found at room temperature, indicating the formation of a second phase as a result of an alteration in the crystalline structure. Polymorphism in this class of materials is in fact quite common and widely reported. In particular, D. G. Billlings and A. Lehmerer reported a phase transition for Lc (Billing and Lemmerer, 2007, 2008; Lemmerer and Billing, 2012), at 240.9 K, from an orthorhombic to a monoclinic phase, hence in reasonable agreement with our optical data. The appearance of a second peak at higher energy is also observed in the case of shorter fully-hydrogenated cations such as butylammonium in the form of n = 1 2D perovskite, (BuA)₂PbI₄ (see Figure S4), with similar phase transition observed in the literature (Billing and Lemmerer, 2007). The 28 nm difference in the peak positions (513–485 nm) is exactly the same for both fully-hydrogenated cations (BuA and Lc). Nevertheless, it is not to exclude a role of the morphology of the sample (thin-film) on the thermodynamics of the phase transition, which can therefore happen at different temperatures or that can result in the co-existence of two segregated phases. In contrast, the fluorinated Lf does not show any additional blue-shifted peak, which can indicate, a different structural arrangement of the material within the temperature range investigated. This indicates that the organic cation does play an ultimate role on the structural and optical response of layered 2D n = 1 HP. Interestingly one may observe that (Lc)₂PbI₄, (Lf)₂PbI₄ and (BuA)₂PbI₄ thin films exhibit very similar optical band gaps at low temperature (Figure 2, Figures S4, S5), thereby suggesting that the distortions of the (Lf)₂PbI₄ perovskite network are similar to the ones of (Lc)₂PbI₄ and (BuA)₂PbI₄ below the phase transition (vide infra). Previous literature about dielectric confinement effect of excitons in PbI₄-based layered semiconductors corroborate the fact that the perovskite undergoes a phase transition upon cooling as a consequence of tuning the length and dielectric constant, for some organic cations (Ishihara et al., 1990; Barman et al., 2003; Naik and Vasudevan, 2010; Straus and Kagan, 2018). These phase transitions can drive changes in band structure that complicate in some cases the evaluation of temperature-dependent optical properties (Even et al., 2015). Surprisingly, a phase transition is absent for some non-fluorinated 2D perovskites such as (C₆H₁₃NH₃)₂PbI₄ (Ishihara et al., 1990) and (PEA)₂PbI₄ (Hong et al., 1992), demonstrating the important influence of the cation on the thermodynamic properties of 2D HP. More recently, N. Kitazawa et al. have indeed systematically demonstrated that very subtle changes for alkyl-based cations in the barrier, may trigger of not abrupt low temperature structural phase transitions including hysteresis (Kitazawa et al., 2011). In addition, Yaffe et al. reported that interactions between 2D perovskites and the environment also affect phase transitions. As an example, they claim that the interaction of an exfoliated sheet with a silica substrate eliminates the freezing/melting phase transition in (BuA)₂PbI₄ (Yaffe et al., 2015), and this distinction must be taken into account when comparing the temperature-dependent optical properties of exfoliated sheets and larger single crystals.

We extended the comparison of the optical properties, by monitoring the emission properties of the fluorinated and non-fluorinated 2D perovskites from r.t. down to 150 K (Figures 2C,D). The emission spectra of (BuA)₂PbI₄ in temperature for comparison, can be found in Figure S5. The results confirm the trend observed in absorption. At room temperature both 2D perovskites present a single emission peak located at 519 nm and 501 nm for (Lc)₂PbI₄ and (Lf)₂PbI₄, respectively. However, the emission response at low temperature presents a second peak at lower wavelength (486 nm, 2.6 eV) for the non-fluorinated cation (Lc) in the range of 250–200 K. In contrast, the emission properties in the case of the fluorinated perovskite (Lf)₂PbI₄, is remarkable different. Only one peak is observed in all range of temperature studied, that indicates the absence of phase transition as a response to the reduction in temperature. Also, the maximum peak in the emission is getting narrower and blue-shifted when the temperature decreases (going from 501 to 482 nm). In both system, notably, no broadband emission is observed at low temperature. In addition, PL decays were monitored, and no appreciable differences were found in the dynamic response of these materials (see Figure S6). In contrast, a notable difference in the photoluminescence quantum yield (PLQY) is observed comparing fluorinated and non-fluorinated 2D perovskite structures. The obtained values fixing the light intensity at 1 sun, are promising and in line to what previously reported 2D perovskites (Cortecchia et al., 2017) being 0.1 % for (Lf)₂PbI₄ and the double, 0.2 % for (Lc)₂PbI₄ (see Figure 3A). As the PLQY is defined as the ratio of the emitted photon flux to the absorbed photon flux, it can be concluded that the 2D perovskite based on non-fluorinated organic spacer presents a more efficient radiative recombination of charges than the fluorinated one. This difference might be related to different film crystalline quality, ultimately associated to the different trap density in the two materials. This observation is in line with the absolute PL intensity behavior with lowering the temperature (see Figure S9). At low T the non-radiative processes and the trapping of charges are slowed down leading to an increase in the radiative efficiency. For the fluorinated sample, the PL peak increases 23-folds from high to low T, while for the non-fluorinated cation the increase is 18 times, indicative for a lower trap density for the latest. This finding highlights the importance of structural engineering, in particular the election of the organic spacer, to control the photoluminescent properties of this class of perovskite.

The energetics of the electrons in the valence and conduction bands represent another key parameter for the technologic exploitation of 2D HP. To have access to the density of occupied electronic states, including the energetics of the valence band maximum (hence the ionization potential, IP), as referred to the energy in the vacuum we performed ultraviolet-photoelectron spectroscopy measurements. The results of UPS measurements for Lc and Lf perovskites are shown in Figure 3B. The valence band (VB) position are extracted using a linear extrapolation of the leading edges in the UPS spectra in agreements with previous works (Gao, 2010; Irfan et al., 2010; Olthof and Meerholz, 2017; Fassl et al., 2018). The measured IP are 5.8 and 5.5 eV for Lc and Lf, respectively and are comparable to the reported IP of CH₃NH₃PbI₃ (Schulz et al., 2014; Fassl et al., 2018). Notably, a change in IP (0.3 eV) was found in presence of the fluorinated Lf spacer. This result is in part surprising, as the inclusion of strongly electronegative atoms in Lf at opposite position to the positive NH3+ group is expected to provide additional negative contribution to the molecular dipole (see also Figures S7, S8). Quantum chemical calculations indeed provide a dipole moment of 23.07 Debye and 32.76 Debye, respectively for Lc and Lf, with the positive side on the NH3+ group. This increase in the dipole moment in turn should result in the up-shift of the electrostatic potential in the vacuum (E_vac), as expressed by the Helmholz relation (Cahen and Kahn, 2003),

(where μ_z is the electric dipole normal to the surface, A is the surface and ε₀ is the dielectric constant) and consequently in the increased of the work function and of the Ionization potential. Periodic DFT calculations performed on slab models further confirm this picture. In Figure 3C, we report the electrostatic potential averaged along the normal of the slab, which clearly highlights the expected up-shift in the vacuum potential, from Equation (1), while in Figure 3D, we report the corresponding electron Density of States (DOS), showing the increase in the IP. We can speculate that this apparent contrast between theory and experiment could be related to non-idealities, as point defects (vacancies or interstitial of cations and anions) or different conformation of the spacers at the surface, which, considering the soft nature of the material and the solution technique employed for its fabrication, are not unlikely. It is also worth to point out that the determination of the IP of hybrid perovskites has always revealed challenging, highlighting a strong sensitivity of this quantity to the substrate (Olthof and Meerholz, 2017), surface defects (Haruyama et al., 2014), and termination (Quarti et al., 2017). Specifically, literature data shows dispersion for the UPS measured IP of the prototypical 3D CH₃NH₃PbI₃ halide perovskite for more than 1 eV (Schnier et al., 2017), which indeed evidences a great sensitivity of the electronic properties of these materials, with respect to the detailed surface structure and synthesis conditions. In this sense, further characterization activity, properly supported by theoretical models, are thus required in order to clarify the mechanisms affecting the energetics of frontier orbitals in hybrid halide perovskites.

4,4,5,5,6,6,7,7,8,8,9,9,9-Tridecafluorononylamine hydroiodide (Lf) was prepared as previously described (García-Benito et al., 2018). Nonylamine hydroiodide (Lc) was prepared by reaction of nonylamine (1 mL, 5.45 mmol) with 57% aqueous HI (0.85 mL, 5.6 mmol) in MeOH (6 mL). After 2 h stirring at rt, the solvent was removed at reduced pressure. The solid material was triturated with Et₂O and Petroleum Ether, filtered and dried under reduced pressure to afford 1.44 g (97% yield) of the ammonium salt.

¹H NMR (400 MHz, CDCl₃) δ 7.56 (s br, 3H), 3.13 (m, 2H), 1.86 (m, 2H), 1.55 – 1.11 (m, 12H), 0.87 (t, J = 6.7 Hz, 3H). ¹³C NMR (101 MHz, CDCl₃) δ 40.58, 31.84, 29.38, 29.22, 28.99, 27.38, 26.67, 22.69, 14.13.

The fabrication of 2D perovskite thin films followed a similar procedure. First, 1.2 M of PbI₂ was dissolved in DMSO at 60°C. Then, the solution was cooling down and 2.4 M of the organic cation was added to the solution. The solution was then deposited onto the substrate via a consecutive two-step spin-coating process at 1,000 rpm for 10 s and 5,000 rpm for 30 s. During the second step, 100 μL of chlorobenzene was deposited. No antisolvent was used to for the 2D perovskite based on Lc cation. The resulting films were then annealed at 100°C for 15 min.

XRD patterns were recorded by X-ray diffractometer (Bruker D8) with Cu kα radiation.

UV-Vis steady-state absorption spectra were acquired with a Perkins Elmer lambda 950 s UV/Vis spectrophotometer using an integrating sphere.

Steady state and time resolved photoluminescence measurement where carried out on Horiba a Fluorolog-3, with a PMT as detector. The excitation source for the TCSPC is a Horiba nanoLED370 with an excitation wavelength of 369 nm, a pulse duration of 1.3 ns and a repetition rate of 100 KHz. PL at different intensity was performed using an Expla PL 2230 laser coupled to a PG400 OPG. The spectra are recorded with a Horiba synapse CCD detector coupled to a monochromator. Steady state and time resolved photoluminescence measurement where carried out on Horiba a Fluorolog-3, with a PMT as detector. The excitation source for the TCSPC is a Horiba nanoLED-370 with an excitation wavelength of 369 nm, a pulse duration of 1.3 ns and a repetition rate of 1 MHz.

Films of 2D perovskites on FTO/glass were transferred into an ultrahigh vacuum (UHV) chamber of the PES system (Thermo Scientific ESCALAB 250Xi) for measurements. UPS measurements were carried out using a double-differentially pumped He discharge lamp (hν = 21.22 eV) with a pass energy of 2 eV and a bias at −5 V.

Perovskite layers for PDS characterization were prepared on spectrosil in an identical way to those on glass/FTO. The samples were mounted in a sample holder filled with Fluorinert FC-770 (IoLiTec) in a nitrogen filled glovebox. A 150 W xenon short-arc lamp (Ushio) provides light for a monochromator (Cornerstone 260 Oriel, FWHM 16 nm) to achieve a chopped, tunable, monochromatic pump beam. The heat caused through absorption of the pump light in the perovskite films changes the refractive index of the Fluorinert. This change is detected by deflecting a probe He–Ne-laser (REO) whose displacement in turn is measured by a position-sensitive-detector (Thorlabs, PDP90A). The magnitude of the deflection is determined by a lock-in amplifier (Amatec SR 7230) and directly correlated to the absorption of the film. To estimate the Urbach energies a python leastsquare routine is used to fit an Urbach tail to the measured absorption spectra in the range of the absorption edge.

Excitation for the PL measurements was performed with a 445 nm CW laser (Insaneware) through an optical fiber into an integrating sphere. The intensity of the laser was adjusted to a 1 sun equivalent intensity by illuminating a 1 cm²-size perovskite solar cell under short-circuit and matching the current density to the J_SC under the sun simulator (22.0 mA/cm² at 100 mWcm⁻², or 1.375 × 10²¹ photons m⁻²s⁻¹). A second optical fiber was used from the output of the integrating sphere to an Andor SR393i-B spectrometer equipped with a silicon CCD camera (DU420A-BR-DD, iDus). The system was calibrated by using a calibrated halogen lamp with specified spectral irradiance, which was shone into to integrating sphere. A spectral correction factor was established to match the spectral output of the detector to the calibrated spectral irradiance of the lamp. The spectral photon density was obtained from the corrected detector signal (spectral irradiance) by division through the photon energy (hf), and the photon numbers of the excitation and emission obtained from numerical integration using Matlab. In a last step, three fluorescent test samples with high specified PLQY (~70%) supplied from Hamamatsu Photonics where measured where the specified value could be accurately reproduced within a small relative error of <5%.

Periodic and molecular electronic structure calculations are performed in the frame of the Density Functional Theory (DFT). Crystal structure relaxation (including both fixed and variable lattice calculations) and band gap estimates are performed using the pseudopotential/plane-wave approach, as implemented in the PWSCF code (Giannozzi et al., 2017). For these calculations, we considered a cutoff of 25 Ry and 200 Ry for the plane-wave expansion of the electronic wavefunction and density, along with ultrasoft pseudopotentials, and a 4 × 4 × 1 mesh of the first Brillouin zone, in the Monkhorst-Pack scheme (the smallest sampling factor is associated to the stacking plane direction, in relation to the corresponding larger parameter in the direct lattice). The exchange-correlation potential employed is cited throughout the text. The dielectric response of the Lc and Lf materials are calculated with the Siesta package (Soler et al., 2001), using the same sampling of the first Brillouin zone, along with double-split quality basis set and PBE functional for the description of the exchange and correlation. We verified the convergence of long-range slab to slab interactions, as reported in the Supporting Information. Finally, we computed molecular properties, i.e., molecular dipole moment, of Lc and Lf using Gaussian16 (Frisch et al., 2016), doting double split 6–31G(d,p) basis set along with PBE exchange correlation functional, for a matter of consistency with the previous calculations. Notably, test calculations with hybrid B3LYP functional provided molecular dipoles in close agreement with those from PBE.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer DC declared a past co-authorship with one of the authors CQ to the handling editor.