Synthesis, characterization and optoelectronic properties of 2D hybrid RPbX₄ semiconductors based on an isomer mixture of hexanediamine-based dications

2.1 Starting materials

The following materials were used as received without any further purification; lead(II) iodide 99% (Sigma-Aldrich 211168), lead(II) bromide 98%+ (Sigma-Aldrich 211141), lead(II) chloride (Alfa 21490), 2-2-4(2,4,4)-trimethyl-1,6-hexanediamine 99% (Sigma-Aldrich 722650), 57% aqueous hydroiodic acid (Sigma-Aldrich 210021), 48% aqueous hydrobromic acid (Alfa-Aesar 14036), 37% aqueous hydrochloric acid (Sigma-Aldrich 320331), anhydrous acetonitrile (Sigma-Aldrich 34851), dimethyl sulfoxide 99.9% (DMSO, Sigma Aldrich 276855) and N,N-dimethylformamide (DMF, Sigma-Aldrich 68-12-2).

2.2 Synthesis of 2,2,4(2,4,4)-trimethyl-1,6-hexanediamine lead halide materials

2.3 X-ray powder diffraction

The X-ray powder diffraction data (XRPD) were obtained from polycrystalline samples at room temperature on a Bruker D8 Advance diffractometer equipped with a LynxEye^® detector and Ni-filtered CuKα radiation. The scanning area covered the 2θ range of 2°–80°, with a scanning angle step size of 0.015° and a time step of 0.161 s. Structural analysis was carried out using the FullProf software [58].

2.4 Scanning electron microscopy and EDX

The SEM images and EDX data were recorded on an EVO-MA 10 Carl Zeiss instrument equipped with a 129 eV resolution INCAx-act Silicon Drift Detector. EDX spectra were acquired at 15 kV and images at 5 kV accelerating voltage. Samples were measured without the application of any conductive coating.

2.5 Optical measurements

The UV/Vis optical absorption (OA) spectra were recorded on a UV-1800 UV Shimadzu spectrophotometer in the range of 200–800 nm, at a sampling step of 0.5 nm using 1.5 nm slits, a combination of halogen and D₂ lamps as sources. The samples were measured as thin or thick spin-coated films on quartz or ITO substrates, after subtracting the substrate’s spectra as reference. The photoluminescence (PL) and photoluminescence excitation (PLE) spectra were obtained from thick deposits on quartz plates, mounted in a Hitachi F-2500 FL spectrophotometer employing a xenon 150 W lamp and an R928 photomultiplier. All the above data were obtained at room temperature. For the measurements at T = 77 K liquid N₂ was used as a blanket to a sealed quartz tube hosting the materials under argon atmosphere.

3.1 Crystal structures

The Powder XRD patterns of m1, m2 and m3 have been indexed, where all compounds crystallize with monoclinic symmetry with strong resemblance to the 2D hybrid semiconductor [NH₃(CH₂)₆NH₃]PbI₄ [59] which is presented in Figure 1, and are presented in Figure 2 along with the latter’s compound computed XRD pattern. No peaks were assigned to unreacted PbCl₂, PbBr₂ or PbI₂. However, due to the low symmetry of the perovskite compounds and the lack of high-quality single crystals, the XRPD patterns are not suitable for a full structural Rietveld analysis to determine the atomic positions.

Figure 1:

Crystal structure of [NH₃(CH₂)₆NH₃]PbI₄ from ref. [59] (COD ID 7203879). Unit cell edges are denoted with dark blue lines, whereas the corner-sharing [PbI₆]^4– octahedra edges are denoted with violet lines.

Figure 2:

Powder XRD patterns with normalized diffraction intensities at room temperature for m1 (green), m2 (red) and m3 (blue) and the computed powder XRD of the compound in Figure 1 (magenta).

Peaks at 2θ of ca. 6° and ca. 12° ± 0.1° are associated with the inorganic interlayer distance of the 2D perovskites according to literature [60], [61], [62] and assigned to the (100) and (200) planes, respectively, in compounds m1, m2 and m3. It is worth observing that the position of these peaks is almost independent from the choice of the halogen atom since the interlayer distance, giving rise to the strongest XRD peak, is mostly affected by the organic spacer and not by the halogen atom. As 2,2,4(2,4,4)-trimethyl-1,6-hexanediamine is significantly bulkier than the 1,6-hexanediamine, larger d-spacings are expected for the (h00) reflections (assuming that a axis is vertical to the inorganic layers), thus, shifting the 2θ angles to lower values than in [NH₃(CH₂)₆NH₃]PbI₄. Very recently, a similar 2D crystal structure was reported for the 2D-(MBA)₂PbI₄ (MBA = chiral R/S-methylbenzylammonium) which produces a similar diffraction pattern to the title compounds [63] as well as a similar work by Ma et al. [64] where chirality further induces more exotic light–matter interaction phenomena.

The amine configuration has been optimized with ab initio dynamics (Supplementary Tables 2 and 3) in order to check for self-consistency in the XRD large plane spacings. The computed average protonated amine lengths plus the halogen-Pb-halogen distances amounts to almost the experimentally deduced d values, as derived from the first low angle peak in the Powder XRD patterns.

In the Supplementary Material, the powder XRD patterns (Supplementary Figures S1, S5 and S9) for the iodine synthesis variations are presented. The low angle peaks presented, indicate in all cases a 2D superstructure. Sample m1_ns3 (Supplementary Figure S5, which is similar to m1 besides the extra HI and PbI₂ but within the ideal stoichiometry) has two sharp XRD peaks at low angles showing the formation of a unit cell with a smaller large axis by ca. 1.4 Å smaller. It is possible that this formation is due to the existence of two different phases derived from the 2,2,4 and 2,4,4 amine packings, which differ as found in the computed ab initio isomer lengths, described in the Supplementary Material section as the distance between the N atoms on each isomer.

The as prepared powders of the materials were stored in the dark at room temperature for a period of three years. The samples were dried before characterizing them. Compound m1 appeared to decompose at high drying temperatures (∼70 °C). The m2 and m3 materials were dehydrated by partial thermal treatment in a desiccator set at 120 °C, after which these were reversed possibly from hydrates to non-hydrates. The color of the powders did not change after the thermal treatment. Figures 3 and 4 present the XRD patterns for compounds m2 and m3, respectively, as prepared and after the aforementioned storage period and the thermal treatment. It is possible that H₂O molecules were partially intercalated in the perovskite structure, so that the organic–inorganic framework was retained, but the X-ray diffraction pattern became more diffuse, as in the case of CH₃NH₃PbI₃ upon hydration [65]. For both compounds the first low angle peak has become more intense and the other peaks seem to be minimized while new peaks have appeared. The normal degradation appears to induce change mostly to the inorganic layers, while the organic spacer is not altered significantly.

Figure 3:

XRD patterns of m2 (a) as prepared and (b) after three years.

Figure 4:

XRD patterns of m3 (a) as prepared and (b) after three years.

3.2 Electron microscopy

SEM images of all three compounds as prepared are presented in Supplementary (Figure S13). In particular, Supplementary Figure S13 (a, b), (c, d) and (e, f) show images of the as prepared compounds m1, m2 and m3, respectively, where some pin holes are observed and a flower like structure, while it is also seen that the anion substitution alters the perovskite morphology. Compound m1 forms a grain-like microstructure while its variants m1_ns2 (Supplementary Figure S2) and m1_ns3 (Supplementary Figure S6) form needles on top of the flower-like microstructure. The morphology of the variant m1_ns4 appears to be plate-like (Supplementary Figure S10). On the contrary, compounds m2 and m3 form needle-like microstructures (Supplementary Figure S13). Should a synthesis proceed with a surplus of the amine and smaller quantity of hydroiodic acid, the material assumes a soft-plastic texture.

Comparison of the SEM images of m2 and m3 are presented further below, before and after the three-year storage period except for m1 since it was the only one that degraded beyond characterization capability. In Figure 5, m2 is presented, where after drying the samples a more well-formed cubic shaped microstructure is observed.

Figure 5:

SEM images of m2 as synthesized (a–c) and dried (d–f).

A comparison of the SEM images of m3 (Figure 6) allowed us to confirm that the morphology is quite similar except for some tetragonal particles observed in the as-synthesized compound (red circles). After drying and overnight storing of the compound under vacuum, the small tetragonal units disappeared and a needle-like microstructure had formed.

Figure 6:

SEM images of m3 as synthesized (a–c) and dried (d–f).

According to EDX measurements (Supplementary Figures S14–S16), all compounds showed a molar ratio Pb:X (X = I, Br, Cl) of 1:4. This is in agreement with the results of the powder XRD measurements (Figure 2) that indicate that the materials form 2D arrangements, where the corner-sharing topology leads to the aforementioned stoichiometry.

3.3 Optical studies

The optoelectronic properties of the compounds were investigated using UV/Vis spectroscopy (Figure 7) at room temperature. The optical absorption (OA) spectra reveal excitonic peaks at 494, 400 and 332 nm for compounds m1, m2 and m3, respectively; for compound m2 the peak is not as strong or sharp. Compound m1 contains iodine which is larger and less electronegative than bromine and chlorine in compounds m2 and m3, respectively, resulting to an excitonic red shift, in agreement with previous studies of similar compounds [59, 66, 67]. In all three compounds, a second peak is observed at higher energies relative to the n = 1 excitonic peak. These low wavelength excitonic peak positions at 318 and 258 nm correspond to the absorption of the PbX₆⁴⁻ standalone units of the analogous Br and Cl compounds [2], respectively, which are, however, not completely isolated but slightly interact with the environment, as for example the isolated PbBr₆⁴⁻ peak would be located at ca. 308 nm [68].

Figure 7:

UV/Vis OA spectra of m1, m2, and m3 as prepared.

The variant m1_ns2 shows excitonic peaks at 487 nm (Supplementary Figure S3), due to the small amount of lead precursor used for this variant; possibly only small sized platelets formed leading to a considerable blue shift size effect. Sample m1_ns4 exhibits peak at 502 nm (Supplementary Figure S11), that are slightly shifted relative to m1, which probably can be attributed to steric effects of the amine and “loose” hydrogen bonding and their modulation on energy band gaps of iodoplumbates [69, 70]. Variant m1_ns3 exhibits excitonic peak at 363 nm (Supplementary Figure S7), as noted before, as well as a weak 2D excitonic peak at 506 nm. This shows that minute changes, i.e. doubling the precursor HI moles, lead to the formation of 0D entities of the PbI₆⁴⁻ type and the minimization of the number of the 2D inorganic sheets.

In Figure 8, the UV/Vis spectra of m2 as prepared and aged samples are presented. The excitonic peak at ca. 400 nm is still present but the peak at 310 nm has strongly decreased, which implies that after the thermal dehydration of the stored compound m2, the peak corresponding to the 0D isolated lead bromide octahedra has been degraded, a phenomenon that has been observed before for similar compounds [71]. According to the SEM images (Figure 5), m2 after its long storage has in fact formed better crystals. Thus, the m2 compound has not been degraded by oxygen and humidity, while compound m3 has been affected as discussed further below.

Figure 8:

UV/VIS OA spectra of m2 as prepared (a) and after three years storage (b).

A peculiar behavior is observed for compound m3 after its prolonged storage. As it can be observed in Figure 9, the peak at 332 nm has been minimized and the peak at 286 nm has been widened and enhanced, both remaining despite the loss of the 257 nm peak. It is possible that the 2D perovskite structure had its superstructure broken down along the long a axis to accommodate both the organic cation and any H₂O molecules trapped in between, and this has partially degraded the inorganic layers [72], [73], [74], since the organic spacers here are partially hydrophilic.

Figure 9:

UV/VIS OA spectra of m3 as prepared (a) and after three years storage (b).

By extrapolating the linear part of the Tauc graphs for the optical absorption, it is possible to get an estimate of the energy band gap values (Supplementary Figure S17) of the as synthesized compounds m1, m2 and m3; these were computed to be 2.82, 3.29 and 3.92 eV, respectively (see Supplementary Material). Obtaining a rough experimental estimate of the energy band gap values, from the first minimum of the OA spectra at higher energies than the excitonic peak, E_g values are estimated at 2.69, 3.28 and 4.0 eV for m1, m2 and m3, respectively, leading to experimental excitonic binding energies of 185, 195 and 253 meV, respectively.

Figures 10–12 present the photoluminescence (PL) spectra of m1, m2 and m3, respectively, while Figures 13 and 14 compare the time variation of the PL of m2 and m3, respectively, after their prolonged storage. In all three compounds, broad bands are observed rather than sharp excitonic peaks, before and after their storage. PL emission spectra exhibit excitonic broad bands as a double peak at 512 and 531 nm for m1, 414 nm for m2 and 346 nm for m3. Comparing the OA and PL emission spectra, the excitonic peaks appear to have an average Stokes shift of 15 nm for all compounds, indicative of only few defect crystalline states of all compounds [75]. The double peak for m1 is possibly attributed to two different existing structural variants, exhibiting, however, the same XRD pattern, or even more probably due to defects. It is interesting that the as prepared m3 shows also a sharp peak at 400 nm and a broad one centered at 465 nm, both latter cannot be linked to other experimental observations of similar materials.

Figure 10:

OA (red) and PL (green) spectra of m1.

Figure 11:

OA (red) and PL (green) spectra of m2.

Figure 12:

OA (red) and PL (green) spectra of m3.

Figure 13:

PL spectra of m2 as prepared (b) and after three years storage (a).

Figure 14:

PLE (left) and PL (right) spectra of m3 as prepared (a) and after three years (b).

As seen in Figure 13, compound m2, presented minor alterations in the photoluminescence peaks after the three years storage period. A more intense and sharper peak has appeared which is slightly blue-shifted to ca. 408 nm in contrast to the as prepared sample, ca. 414 nm, yet both spectra are quite wide. The PL spectra of the as prepared sample has been enhanced in order to be comparable to the aged one, which are comparable due to the similar geometry used for both.

For compound m3, the PL spectra have revealed major differences after the three-year storage period (Figure 14). According to the OA spectra, m3 retains the 2D excitonic peaks, but the as prepared sample exhibits the PL peak at 347 nm, while the stored sample exhibits a broad and red-shifted excitonic PL peak to 363 nm, due to defects, and at the same time the 400 nm PL peak has disappeared. The PLE spectra, however, which should resemble the OA spectra, show that the spectra of both aged and fresh m3 have a strong peak at 369 nm, probably denoting some defect states responsible for the low energy emission at 550 nm.

In Figure 15, 3D PL spectra of compound m2 after being exposed to humidity and oxygen are presented. A blue up to green luminescence is observed that may be attributed to defects in the crystal structure in accordance to the data of Figure 13. According to Figure 6 the SEM images show more well-formed crystals with more sharp edges after the samples were exposed to humidity, as compared to the as synthesized sample which presents a smoother surface.

Figure 15:

3D PL spectra of m2 at room temperature after three years of storage.

In Figure 16, 3D PL spectra of m3 after being exposed to humidity and air are presented. A broad blueish-green luminescence is observed for a variety of excitation wavelengths from 260 nm up to 400 nm, evident of white like luminescence.

Figure 16:

3D PL spectra of m3 at room temperature after three years of storage.

In the Supplementary Material, PL and PLE spectra of m1_ns2, m2 and m3 are provided, measured at 77 K for the as prepared compounds. These measurements were performed to investigate phenomena responsible for the low PL intensities observed compared to similar 2D lead halide perovskites. The PL spectra at 77 K of some of m1_ns2 show a single peak at 509 nm (Supplementary Figure S18), slightly blue-shifted in comparison to the RT spectra (Supplementary Figure S8). These may be due to some structural variation, while the PLE spectra agree with the RT OA spectra as far as the exciton is concerned. The 77 K spectra of m1_ns2 also display a shoulder at ca. 600 nm, probably due to complex defect states linked to the Pb⁺² ion in a halogen environment, as for example seen in ref. [76]. Compound m2 shows a peak at ca. 410 nm in the RT spectra (Figure 11) as well as at 77 K, however, the PLE spectra at 77 K reveals a characteristic absorption at 300 nm (Supplementary Figure S19) as well as a 368 nm PLE peak. The 300 nm PLE peak was used to collect a PL spectra with λ_exc = 300 nm, that shows that at 77 K this excitation yields a broad exciton peak centered at 407 nm as observed in Supplementary Figure S19. Exciting m2 with higher or lower energies, i.e. 280 or 320 nm, provides PL spectra with less intensity than using 300 nm excitation. It is possible that the degraded m2 has quantum confined sheets that instead of showing absorption at ca. 330 nm they have blue shifted at lower wavelengths, thus, using 300 nm radiation allows the broad 2D excitonic emission by virtue energy transfer. It is suggested that 300 nm radiation induces photoluminescence at high energies which is transferred to the remaining un-degraded 2D sheets to be emitted mainly as 407 nm radiation. In this process, all other radiative processes yield the broad PL peak of 407 nm. All these elucidate the complexity and role of the defects and multi-structural diversity of the lead halide perovskites. It appears that this perovskite could emit broad UV-blue light if excited with a GaN LED or electrically activated.

Finally, m3 shows an excitation dependent PL spectrum at 77 K (Supplementary Figure S20), centered far from the exciton peak at 500–560 nm, while the PLE spectrum reveals that absorptions at 303, 333, 368 nm appear to be responsible for this wide low temperature PL peak. It is conjectured that this PL behavior is due to a mechanism with which the exciton formed by the high energy excitation transfers its energy to defects of lower energy or transfer it towards the organic–inorganic artifacts which could form by lattice interruption introduced by the cooling process. Overall, it can be assessed that these types of materials, which deviate from the standard LD HOIS perovskites, have a quite complex behavior. Finally, it is important to note the similarity of the room temperature absorption spectra for the 2D materials in this work and those in refs. [63] and [64] based on RMX₄ which are broad rather than sharp as it usual observed in 2D lattices. It appears, however, that the use of the mixture of 2,2,4 and 2,4,4 isomers as well as including their enantiomers provide a sharper exciton peak at 494 nm.