Synthesis, Characterization, and Stability Studies of Ge-Based Perovskites of Controllable Mixed Cation Composition, Produced with an Ambient Surfactant-Free Approach.

In this report, we have applied a facile, ligand-free, ambient synthesis protocol toward the fabrication of not only a series of lead-free Ge-based perovskites with the general formulation of MA1-x FA x GeI3 (where x was changed from 0, 0.25, 0.5, 0.75, to 1) but also CsGeI3. Specifically, our methodology for producing ABX3 systems is generalizable, regardless of the identity of either the A site cation or the X site halide ion. Moreover, it incorporates many advantages, including (i) the possibility of efficiently generating pure Ge-based perovskite particles of any desired chemical composition, (ii) the use of readily available, commercial precursors and comparatively lower toxicity solvents, (iii) the practicality of scale up, and (iv) the elimination of the need for any superfluous organic surface ligands or surfactants. In addition to providing mechanistic insights into their formation, we have examined the chemical composition, crystallite size, morphology, surface attributes, oxidation states, and optical properties of our as-prepared perovskites using a combination of diffraction, microscopy, and spectroscopy techniques. Specifically, we noted that the optical band gap could be reliably tuned as a function of chemical composition, via the identity of the A site cation. Moreover, we have probed their stability, not only under standard storage conditions but also, for the first time, when subjected to both e-beam- and X-ray-induced degradation, using cumulative data from sources such as synchrotron-based scanning hard X-ray microscopy. Importantly, of relevance for the potential practical incorporation of these Pb-free perovskites, our work has emphasized the possibility of controlling the chemical composition within Ge-based perovskites as a means of rationally tuning their observed band gaps and optical behavior.

Introduction

Solar cell technology is an attractive, promising avenue of research, in part because sunlight is free and abundant, and unlike fossil fuels is fundamentally renewable and clean.[1,2] In this context, hybrid organic–inorganic halide perovskites have shown very promising performance for energy conversion applications.[3] Several factors appear to play a role in this behavior, with one of the most important ones being ascribed to the capability of tuning band gaps[4] through externally controlled variables such as chemical composition. However, traditional Pb-based perovskites are relatively unstable in the presence of air and moisture.[5] Hence, the major challenge has been to seek viable and practical alternatives to Pb without necessarily compromising upon the intrinsically high efficiency and favorable photophysical properties of this class of materials. There are many potential chemically diverse solutions to this problem, which we discuss in detail in the Supporting Information.

Nonetheless, we are primarily focused on Ge-based perovskites herein. What is our motivation? Ge-based perovskites possess an attractive direct band gap of 1.88 eV, which is comparable in magnitude with the 1.48 eV value reported for conventional Pb-based perovskites. Moreover, as compared with elements such as Pb, Sn, Cd, and Be, which possess similarly narrow band gaps, Ge is less toxic and more earth-abundant. Therefore, in light of these considerations, our approach in this study herein has been to substitute Pb with Ge.

Based on the prior literature, it is evident that Ge-based perovskites are especially promising to consider because of their following advantages.

Ge-based perovskites maintain a proper, direct band gap, which can be tuned by changing the A and the X sites. Specifically, CsGeI3, MAGeI3, and FAGeI3 evince experimentally measured band gaps of 1.6, 1.9, and 2.2 eV,[6] respectively, while CsGeCl3 and CsGeBr3 possess corresponding band gaps of 3.67 and 2.32 eV,[7,8] respectively. Indeed, with band gaps capable of being tailored to values lower than 2.0 eV, Ge-based perovskites have the potential to absorb light energy with wavelengths smaller than 610 nm (i.e., orange).

As compared with Sn-based perovskites, theoretical calculations suggest that Ge-based perovskites are characterized by a more negative (i.e., more favorable) free energy of formation, thereby implying a greater stability of MAGeI3 versus MASnI3, for instance.[9]

Ge-based perovskites possess high ionic conductivity[7,10] and good electronic properties.[7,11]

Ge is earth-abundant and evokes comparatively little toxicity or environmental concerns, thereby overcoming a major drawback of either Pb or Sn-based perovskites.

Ge is much lighter than both Pb and Sn, a factor which should improve upon the power density of Ge-containing materials. Overall, Ge-based perovskites deserve further attention and effort.

It is worth pointing out that the majority of work on Ge-based perovskites is based on theoretical considerations of their stability, band gap, orbital charge density, electronic character, and second-harmonic generation properties. Fewer papers regarding their experimental synthesis exist. Many of these studies either employ complicated methods or have limited practical applicability. In the initial synthesis of hybrid Ge iodide perovskites, for example,[6] a rather complicated hot coprecipitation method was utilized in the presence of H3PO2 and concentrated HI, which are both relatively hazardous; in addition, H3PO2 resides in the U.S. Drug Enforcement Agency list of controlled substances. Although this procedure has subsequently been utilized to produce Ge-iodide perovskites incorporating different representative A site cations, including Cs+, MA+, FA+, CH3C(NH2)2+, C(NH2)3+, (CH3)3NH+, and (CH3)2CHNH3+, the reaction remains a multistep process, requiring elevated operating temperatures and an inert reaction atmosphere. Moreover, the isolated morphology consists of inhomogeneous, micron-sized crystals. A second strategy was associated[12] with combining MAI, MABr, and GeI2 precursors together in dry DMF followed by a reaction within an inert atmosphere at room temperature. Though inherently simpler, only MAGeBrI3– could be readily synthesized in this process, which nevertheless took more than 24 h to complete. A third protocol, focused on producing CsGeI3 colloidal nanocrystals alone,[13] used a hot-injection technique, relying on relatively high temperatures, and necessitated operation under vacuum.

Hence, given this relatively limited existing body of prior work, we have sought to develop a generalized and flexible way of generating Ge-based perovskites, preferably using a relatively simple, room-temperature reaction, which could be meaningfully expanded to fabricating more than one type of Ge-based perovskite. Specifically, in this study, we have modified and adapted traditional ligand-assisted reprecipitation methods for Pb-based perovskite synthesis[14,15] with notable variations. Indeed, we provide a facile and efficient synthesis method that in theory can be applied to families of Ge-based perovskites incorporating different cation A sites and anion X sites, so as to generate a range of important candidate materials, such as but not limited to cesium germanium iodide, methylammonium (MA+) germanium iodide, and formamidinium (FA+) germanium iodide. Moreover, our protocol is capable of enabling the synthesis of more compositionally variable series of perovskites, such as MA1–FAGeI3 (with x altering from 0, 0.25, 0.5, 0.75, to 1) in addition to MA/FAGeBrI3– (with x = 1 and 2).

Significantly, we could tune the optical performance, especially the band gap, of our materials by creating a mixed MA+ and FA+ cation Ge-based perovskite in which the relative ratios between the two cations were systematically varied. By analyzing modifications in our X-ray diffraction (XRD) patterns and photoluminescence (PL) spectra, we concluded that, indeed, we could systematically tune and presumably optimize the chemical compositions and the band gaps of not only our Ge-based perovskites in general but also, more specifically, our family of mixed cation Ge-based perovskites. As evidence of this accomplishment, we were able to produce a series of Ge perovskites, possessing a well-defined range of different discrete colors and absorption wavelengths.

To summarize the extent of the contributions emanating from our efforts, we have synthesized and characterized (i.e., through the prism of tailorable parameters, such as but not limited to morphology, monodispersity, and chemical purity) a number of Ge-based perovskites, prepared using a relatively facile, reasonable, and generalizable synthesis process. In particular, we have analyzed (i) morphology through the mediation of transmission electron microscopy (TEM), scanning electron microscopy (SEM), and high-resolution transmission electron microscopy (HRTEM), (ii) chemical composition and stability using a combination of XRD, scanning hard X-ray microscopy (SHXM), and X-ray photoelectron spectroscopy (XPS), (iii) surface chemistry with Fourier Transform infrared (FT-IR) spectroscopy and XPS, and (iv) optical properties using UV–visible (UV−vis) and photoluminescence (PL) spectroscopies. In so doing, we have achieved structure–property relationships that correlate chemical composition (especially with respect to stoichiometrically mixed cation perovskites) with optical properties, for example. Moreover, we have investigated the key roles of solvent–precursor interactions in the context of acquiring useful understanding of the growth mechanism of these materials. Indeed, it is worth emphasizing that the focus of our paper is on investigating the chemistry of the materials themselves as opposed to measuring their device characteristics, which is beyond the scope of the present work.

We were particularly interested in analyzing the evolution of physical structure and chemical composition as a function of varying irradiation conditions. The results of this series of stability tests under ambient environment and inert atmosphere conditions attest to their degradation behavior and potential for long-term use in optoelectronic applications. Electron-beam-induced degradation under high energy electron irradiation conditions has been previously observed not only for Pb-based perovskites[16] but also for CsGeI3.[13] It should be emphasized that, whereas the corresponding X-ray-induced degradation has been noted and studied for Pb-based perovskites, comparable experiments have yet to be carried out for their Ge-based analogues.

As such, we undertook similar types of experiments herein. To highlight the comprehensive nature of our efforts to understand the potential susceptibility to radiation of our Ge-based perovskites, we probed their stability and durability through a series of X-ray-based measurements spanning from the macroscopic to nanoscopic range, involving X-ray diffraction, X-ray photoelectron spectroscopy, and scanning X-ray microscopy studies, incorporating high-resolution 2D elemental and morphological mapping. These complementary X-ray-based analyses seek to generate data on morphology, crystallite structure, surface properties, and elemental distribution within our samples. By means of comparison, we have also analyzed the stability of these materials, when subjected to analogous complementary electron-beam-based irradiation associated with TEM, HRTEM, and SEM. Moreover, as a relatively new characterization technique,[17−21] synchrotron-based SHXM imaging with a sub-15 nm probe size has not been previously applied as a means of spatially identifying bulk and surface distributions of different elements within solar perovskite-based materials. Our reported data therefore represent a demonstration of principle for the promising potential of this unique characterization method.

Results and Discussion

Composition and Structural Characterization

By following a facile synthesis procedure, we were able, with relative ease, to generate a series of Ge-based perovskites with variable A site cation occupancy. Hence, in the absence of any organic ligand, a series of Ge-based perovskites, including MA1–FAGeI3 with x = 0.25, 0.5, and 0.75; MAGeI3 (x = 0); FAGeI3 (x = 1); and CsGeI3, were readily synthesized. We found that, with different A+ cations, the isolated perovskites evinced a corresponding set of varied coloration. Specifically, representative samples of MAGeI3, FAGeI3, and CsGeI3 were dark red, yellow, and black, respectively, in color. Furthermore, within the series of MA1–FAGeI3 with x = 0.25, 0.5, and 0.75, the as-generated perovskites possessed individual characteristic hues and became more yellow with increasing x values, or conversely, decreasing MA+ content (Figure S1). As such, our studies demonstrated that the identity of the A site cation could indeed alter band gaps, affect the amount of light absorbed, and ultimately determine the observed color of as-prepared perovskites.

To confirm the formation of our as-prepared samples, we obtained XRD patterns so as to verify the crystal structure and crystallinity, as shown in Figures and 2. All germanium perovskites, including CsGeI3, MAGeI3, and FAGeI3, ought to exhibit a trigonal structure with a space group of R3m. Based upon previously reported XRD patterns of MAGeI3 and FAGeI3 (Figure A,G), our data (Figure B,F) are consistent with this prior study.[6] The most intense peaks located at 14.79, 25.97, and 29.58° for MAGeI3 could be potentially assigned to the (101), (20-1), and (202) planes, respectively. In addition, peaks situated at 14.37, 25.62, and 28.9° for FAGeI3 could be identified with their (101), (20-1), and (202) planes, respectively.

Figure 1

Collected XRD patterns of as-prepared (B) MAGeI3 (black), (C) MA0.75FA0.25GeI3 (red), (D) MA0.5FA0.5GeI3 (blue), (E) MA0.25FA0.75GeI3 (magenta), and (F) FAGeI3 (olive) by comparison with (A) the standard diffraction pattern for MAGeI3 (black) and (G) the standard diffraction pattern for FAGeI3 (olive). Peaks are labeled with the presumed plane orientation. The yellow square highlights the area associated with the most intense (202) peak, a region that is magnified and shown in detail. Red arrows point to the (10-5) peak for each sample, whereas the blue arrows can be ascribed to the (220) peak.

Figure 2

(A) Standard diffraction pattern (red) for CsGeI3 and (B) the collected XRD pattern (black) of as-prepared CsGeI3. The peak labeled with a star in panel (B) can be assigned to CsI impurities.

Furthermore, XRD patterns associated with MA1–FAGeI3 with x = 0.25, 0.5, and 0.75, respectively (Figure C–E) yield features with similar profile shapes to those of the parent compounds of MAGeI3 and FAGeI3, respectively. However, the peaks within the as-collected XRD patterns do shift toward lower angles with increasing x. The trend is particularly evident when analyzing magnifications of the XRD patterns (obtained at an energy of 8.04 keV, corresponding to Cu Kα radiation) in the region from 25 to 30° (Figure S2). In that example, the most intense peak of the (202) plane shifted from 29.58, 29.44, 29.19, 29.14, and ultimately to 28.9° with increasing x. Nonetheless, in Figure B–F, we noted that the peak located at 43.02°, corresponding to, the (10-5) plane (designated by the red arrow), reduced in size with increasing x values (i.e., lower MA+ content) to 0.5, relative to the peak associated with the (220) plane (highlighted by the blue arrow). Indeed, the (10-5) peak intensity keeps decreasing with a corresponding increase in x, thereby suggesting that the crystal structure alters upon substitution at the A+ cation site.

With respect to MAGeI3 and FAGeI3, based upon Bragg’s equation, whereas the peak shifted to lower angles, the corresponding interplanar distance d likely increased, indicative of crystal lattice expansion. Moreover, the FA+ cation is larger in size than the MA+ cation, again consistent with the trend in the peak shift noted within the diffraction patterns. Hence, assuming reasonable sample crystallinity, we calculated the crystallite size for each of our series of MA1–FAGeI3 perovskites (with x = 0, 0.25, 0.5, 0.75, and 1) to be 45.7, 24.4, 36.4, 39.5, and 52.6 nm, respectively, using the Debye–Scherrer equation. Hence, upon mixing MA+ with FA+ cations, the apparent crystallite size was reduced.

It is also worth noting that the XRD pattern of our as-prepared CsGeI3 also corresponds well with a reported database pattern (Figure .[22] However, because CsGeI3 is relatively unstable in air and easily decomposes into its constituent CsI and GeI2 building blocks, we found that our isolated XRD pattern showed a tiny peak located at around 27.72° (labeled by a star), which could be likely ascribed to CsI. The most intense peaks are situated at 25.2, 26.0, and 29.9° and correspond to the (003), (20-1), and (202) planes, respectively. Moreover, the computed crystallite size for CsGeI3 is 57.6 nm, which is comparable in magnitude with those of both MAGeI3 and FAGeI3. This result highlights that, in general, the mixed cation perovskites yield a relatively lower crystallite size (i.e., 24.4, 36.4, and 39.5 nm) than the parent single cation perovskites of MAGeI3, FAGeI3, and CsGeI3, respectively (i.e., 45.7, 52.6, and 57.6 nm).

We subsequently characterized the corresponding morphologies of these as-prepared samples, using both SEM and TEM analysis (Figure . From the SEM images (Figure A,D,G,J,M), we noted that the MA1–FAGeI3 (x = 0, 0.25, 0.5, 0.75, and 1) samples yielded composition-dependent morphologies in terms of size and shape. Specifically, when x = 0, the SEM image (Figure suggests that the MAGeI3 sample maintains a cubic shape with a roughened external surface. The sizes of the cubic particles measured ∼1.488 ± 0.468 μm. The corresponding TEM data were consistent with and corroborated the SEM image of cubes, nominally characterized by a coarse outer layer. The presence of a roughened surface, as noted in both the SEM and TEM images, coupled with the observation of a solid inner core by TEM, indicates that (i) the roughened surface is likely a product of the air-sensitive decomposition process (Section ) and that (ii) the as-generated micron-scale cubes are not porous.

Figure 3

SEM images (A,D,G,J,M,P), TEM images (B,E,H,K,N,Q), and high-resolution TEM images (C,F,I,L,O,R) of as-prepared samples: MAGeI3 (A–C), MA0.75FA0.25GeI3 (D–F), MA0.5FA0.5GeI3 (G–I), MA0.25FA0.75GeI3 (J–L), FAGeI3 (M–O), and CsGeI3 (P–R). The inset image in panel (F) represents the FFT image of MA0.75FA0.25GeI3. The red lines with the arrows highlight indices of the Ge-containing perovskite within each HRTEM image. In panel (R), the green circle can be ascribed to CsI, whereas the yellow circle can be assigned to amorphous Ge.

Upon the incorporation of more FA+ cation content, the edges of the discrete and reasonably well-defined cubic shape were systematically blunted, eventually yielding a rounder, more circular structure. Indeed, the SEM images for MA1–FAGeI3 (x = 0.25, 0.5, 0.75, and 1) suggest that, with increasing x values, the samples evolve into more irregularly shaped particulate motifs, possessing average sizes in the range of 1.337 ± 0.363, 2.311 ± 0.747, 1.932 ± 0.972, and 1.641 ± 0.639 μm, respectively. Upon the addition of a second cation, the sizes of the particles increased in magnitude and the isolated size distribution widened, as the particles became more polydisperse in nature. The corresponding TEM images (Figure B,E,H,K,N) corroborated the presence of a crystalline particulate motif, in agreement with the SEM data.

HRTEM images (Figure C,F,I,L,O) were acquired to provide more nuanced insights into the structure of our Ge-based perovskites. To the best of our knowledge, no HRTEM data have been reported on organic-cation-containing Ge-based perovskites to date. From the HRTEM images of individual particles of MAGeI3, MA0.75FA0.25GeI3, MA0.5FA0.5GeI3, MA0.25FA0.75GeI3, and FAGeI3, all of these five samples appear to possess a high degree of crystallinity, regardless of chemical composition, with measured lattice d spacing values of 0.356, 0.362, 0.386, 0.394, and 0.395 nm, respectively. An FFT image of the diffraction pattern of one of the products has been inserted as an inset to Figure F so as to highlight the high crystallinity of our product. The 0.356 nm TEM-based lattice d spacing corresponds to the MAGeI3 (20-1) plane with a calculated d spacing of 0.352 nm, based upon the indexing of our XRD pattern. Moreover, the TEM-based d spacing of FAGeI3 with 0.395 nm is in agreement with the XRD-derived d spacing associated with the (003) plane of 0.391 nm. The systematic increase in the measured lattice d spacing values as a function of chemical composition (i.e., increasing x in MA1–FAGeI3) is consistent with the corresponding incorporation and facet exposure of more FA+ cationic content within the underlying MAGeI3 lattice.

With respect to CsGeI3 in particular, it evinced a particulate morphology with smaller, measured sizes in the range of 263 ± 100 nm (Figure , as corroborated by the corresponding TEM image (Figure . In particular, an evident amorphous feature was noted especially on its surface, coupled with a polycrystalline region near the center of the particle, an observation we can ascribe to an electron-beam-induced degradation of CsGeI3 to its GeI2, Ge, and CsI precursors. This observation will be further discussed in Section in the context of XPS and SHXM results. As shown in the HRTEM image in Figure R, the yellow circle highlights an amorphous surface region, which can likely be attributed to the formation of Ge0, created upon interaction with irradiation. By contrast, the green circle can be ascribed to an area with larger measured d spacings of ∼0.323 and 0.285 nm, which can be assigned to the CsI (110) plane and CsGeI3 (202) plane, respectively, of the XRD pattern (Figure B). Overall, these high-resolution lattice data further correlate with and corroborate our XRD pattern results, which demonstrate that we have been definitely able to prepare pure and highly crystalline Ge-based perovskites.

Surface Characterization (IR and XPS)

In terms of better understanding the surface properties and chemistry of our different Ge-based perovskites, a series of characterization techniques were applied onto these samples, including FT-IR spectroscopy (Figure S3) and XPS (Figures and ). We noted that the presence and identity of exposed surface functional groups are important, as these help to determine the nature of exciton transport, especially with respect to electron and hole transfer toward relevant adjacent layers in direct contact with these materials. If the surface were to be capped by either organic surfactants or relatively insulating ligands possessing long alkyl chains, electron transfer efficiency in addition to conductivity would be reduced. Methylammonium and formamidinium cations both incorporate organic groups. The IR spectrum was taken to not only confirm the chemical compositions of as-prepared samples but also provide information about the nature of the surface pendant groups. Overall, these spectra suggested that the mixed MA1–FAGeI3 perovskites had indeed formed as a stable entity and that the two cations were well incorporated within the underlying lattice. With respect to CsGeI3, we did not find evidence for either the presence of organic ligands or pendant hydroxyl groups adsorbed on its surface, all of which were consistent with expectations.

Figure 4

XPS spectra of as-prepared MAFA1–GeI3 samples: (A–C) MAGeI3; (D–F) MA0.75FA0.25GeI3; (G–I) MA0.5FA0.5GeI3; (J–L) MA0.25FA0.75GeI3, and (M–O) FAGeI3, with different energy regions highlighted. Peaks are associated with (A,D,G,J,M) Ge 2p, (B,E,H,K,N) Ge 3d, and (C,F,I,L,O) I 3d.

Figure 5

XPS spectra of an as-prepared CsGeI3 sample, associated with different elemental regions: (A) Cs 3d; (B) I 3d; (C) Ge 2p; (D) Ge 3d.

The surface chemistries of these perovskites were further analyzed using XPS. Initially, survey spectra were collected to confirm the purity of as-prepared samples (Figure S4). In Figure , magnified areas of specific peak regions corresponding to different discrete elements have been plotted for samples of MA1–FAGeI3 with x = 0, 0.25, 0.5, 0.75, and 1. Figure A,D,G,J,M corresponds to data for the Ge 2p orbital. Figure B,E,H,K,N is indicative of results for the Ge 3d orbital. Figure C,F,I,L,O denotes data associated with the I 3d orbital.

It is worth noting that, for these mixed perovskites, within the Ge 3d region of the spectrum, two peaks were noted in the region from 28 to 34 eV and these could be assigned to Ge0 3d and Ge2+ 3d, respectively; the coexistence of both oxidation states suggests that Ge2+ was being reduced to Ge under X-ray irradiation. This scenario was also reported for Pb-containing perovskites characterized using XPS. Specifically, with longer X-ray exposure times, it was noted that more Pb2+ species were found to have reduced to Pb0.[16] A similar phenomenon was detected under hard X-ray nanoprobe irradiation conditions, as will be discussed later. Furthermore, with respect to MA1–FAGeI3 samples as a function of increasing x, with x = 0, 0.25, 0.5, 0.75, and 1, the Ge2+ 3d peak (shown in blue) shifted in energy from 31.4, 32.67, 33.38, 33.52, and 34.51 eV, as x correspondingly increased from 0, 0.25, 0.5, 0.75, and finally to 1. All spectra were calibrated with respect to the O peak located at 531.0 eV, whose somewhat unexpected presence may have arisen from absorbed O2 upon exposure to air.

Interestingly, we found that the peak area ratio between the Ge2+ and Ge0 3d peaks varied considerably as a function of X-ray exposure time, a finding which is consistent with the results of a previous analogous study on Pb-containing perovskites.[16] The corresponding peak area ratio between the Ge2+ and Ge0 2p peaks (i.e., olive and pink colors, indicative of satellite peaks present) also evinced a similar behavioral pattern. In particular, there is a clear trend regarding the Ge 3d peak position for both Ge0 and Ge2+ because, as x increases, the peak position shifts to a higher energy. This observation suggests that, with greater FA+ cation content, the corresponding electron density surrounding Ge will be lower. In other words, FA+ and MA+ behave perceptibly differently with respect to Ge. That is, MA+ maintains a reduced electron affinity as compared with FA+ with respect to the adjacent Ge-I3– complex.

As for the I 3d region, all samples gave rise to I 3d3/2 and 3d5/2 peaks, consistent with only a single oxidation state. While analyzing and comparing the peak areas and sensitivity factors associated with these two elements, we obtained relative elemental ratios between Ge and I of 1:3.5 for x = 0, 1:3.2 for x = 0.25, 1:3.7 for x = 0.5, 1:3 for x = 0.75, and 1:3.6 for x = 1. These results are all within the experimental error of the actual expected ratio between Ge and I of 1:3.

In Figure , we performed an identical XPS analysis for CsGeI3, which has no C but contains Cs instead. We found that Cs yielded a similar Ge reduction behavior under X-ray irradiation, with the Ge 3d region highlighting two peaks, located at 38.06 and 35.20 eV, which could be ascribed to Ge0 and Ge2+, respectively. In addition, the I 3d region was composed of two peaks, designated as I 3d5/2 and 3d3/2, denoting findings indicative of only one oxidation state. The Cs 3d area also included two signals that could be ascribed to 3d5/2 and 3d3/2. By performing a quantitative analysis, the elemental ratio between Cs:Ge:I was found to be 1:1.1:2.9, a result which closely approximated the expected formulaic stoichiometric ratio of our desired perovskites.

To summarize, the XPS results do not show the presence of an obvious impurity element. In fact, these cumulative data corroborate our previous XRD and IR findings, indicative of perovskites with the correct expected chemical composition and oxidation states with no extraneous surface ligands present. In essence, the Ge2+ 3d peak position varies in a predictable manner, consistent with corresponding variations in the FA+ cation content x. However, what is certainly worth noting is that the XPS measurement process itself (i.e., under X-ray irradiation) likely tended to reduce the amount of Ge present, as manifested in a decrease in the Ge0 3d peak intensity. In effect, the perovskite invariably decomposed, a phenomenon we will discuss in further detail in Section .

Optical Performance (PL and UV–Visible Spectroscopies)

The UV–visible spectra were collected to further investigate the optical properties of our as-prepared Ge-based perovskites, as shown in Figure A. In this figure, we show five spectra, corresponding to our five MA1–FAGeI3 solid-state samples, wherein x = 0, 0.25, 0.5, 0.75, and 1. Upon comparing these five spectra, it is evident that there is a blue shift of the band edge with decreasing MA/increasing FA content. The movements of the positions of these absorption edges correspond well to the observed color changes of these five samples in progressing from red to yellow (i.e., MAGeI3 to FAGeI3). These data also indicate that there might be a corresponding and consistent (though potentially nonlinear) alteration in the measured band gap as a function of chemical composition, an issue which was further explored with accompanying photoluminescence (PL) data. By calculating a series of corresponding associated Tauc plot curves (Figure , our data indicate that systematically changing chemical composition, as defined by the magnitude of x from 0, 0.25, 0.5, 0.75, and finally to 1 within a series of MA1–FAGeI3 samples, can correspondingly tune the band gap from 1.90, 2.03, 2.10, 2.19, and finally to 2.29 eV. Overall, the band gaps could be increased by introducing more FA+ within the structure, thereby corroborating our initial idea and concomitant novelty of our work that the band gaps of Ge-based perovskites could be rationally tuned and optimized through predictive variation of the A site cation identity.

Figure 6

Optical band gap measurements. UV–visible spectra (A) and corresponding Tauc plots (B) of as-prepared samples of MAGeI3 (black curve), MA0.75FA0.25GeI3 (red curve), MA0.5FA0.5GeI3 (blue curve), MA0.25FA0.75GeI3 (pink curve), and FAGeI3 (green curve).

In addition, we acquired a PL emission spectrum using an excitation wavelength at 350 nm with the results presented in Figure . From the curves herein, we noted that there is an obvious trend for the PL emission peak position with increasing FA+ cation amount x. In particular, the peak position for MA1–FAGeI3 could be systematically tailored from 626, 606, 593, 570, and finally to 548 nm by altering the corresponding value of x from 0, 0.25, 0.5, 0.75, and eventually to 1. With higher FA+ incorporation (i.e., larger x), the major PL peak appears to have experienced a blue shift toward lower wavelengths. Moreover, based upon the PL emission of the Ge-based perovskite, we can correspondingly calculate the band gap energy of our materials, and the results herein are consistent with the band gaps reported from the literature,[23] which we have previously discussed in the Introduction. To underscore the connections between the UV–visible absorption and the photoluminescence emission data, we have plotted the two characterization curves together for each sample in Figure S5.

Figure 7

Normalized photoluminescence spectra of MAGeI3 (black curve), MA0.75FA0.25GeI3 (red curve), MA0.5FA0.5GeI3 (blue curve), MA0.25FA0.75GeI3 (pink curve), and FAGeI3 (green curve). All five of these samples were acquired with an excitation wavelength of 350 nm with a 420 nm long-pass filter.

Overall, to summarize our efforts in probing the optical properties of our as-prepared Ge-based perovskites, we acquired UV–visible absorption and photoluminescence spectral data. The Ge-based perovskites evinced an absorption edge located in the visible light range from 520 to 620 nm. The photoluminescence profile of the Ge-based perovskites yielded emission peaks in the range of 548 to 626 nm with corresponding measured band gaps spanning from 2.26 to 1.98 eV. That is, to highlight structure–property correlations, we found that we could tune the observed band gaps of Ge-based perovskites through systematic variations in FA+ content from 0 to 1 within MA1–FAGeI3.

Solvent Effect and Growth Mechanism

We selected representative solvents (i.e., acetonitrile, tetrahydrofuran (THF), ethanol, and dimethylformamide (DMF)), which have been commonly used in the conventional synthesis of Pb perovskites. Upon dissolving the MAI and GeI2 precursors within the solvents, variously colored solutions were obtained. Specifically, a clear yellow solution was generated in the presence of anhydrous ethanol. However, when THF and acetonitrile were used, dark red solutions were achieved. In the case of DMF, a clear yellow solution, similar to that observed when using anhydrous ethanol, was isolated with a high concentration of GeI2; however, upon dilution of this solution, we noted that it became clear and colorless, which was necessary to accurately measure its resulting UV–visible absorption spectral profile. A series of different antisolvents, based on expected product miscibility within these poor solvent additives, were introduced to precipitate the perovskites from solution. Hence, hexane was used as the antisolvent for ethanol, DMF, and THF, whereas toluene was utilized in the presence of acetonitrile.

Upon addition of the antisolvent, a precipitate was indeed formed from each solution, and the resulting product in every case was subsequently characterized by XRD. As such, with ethanol, we isolated a red solid, which was later determined to be pure MAGeI3 (i.e., its trigonal phase). With THF, we collected an orange precipitate that we hypothesize may consist of either a combination of MAI and GeI2 or possibly MAGeI3·xH2O, as it is difficult to clearly differentiate between these two possibilities. When employing DMF as the solvent, as implied, concentration plays a crucial role. Specifically, when the GeI2 concentration was 0.5 M, we generated a red precipitate of MAGeI3, whereas when the GeI2 concentration was 0.1 M, no precipitate perceptibly formed. Finally, with acetonitrile, a yellow-white precipitate was produced of the initial MAI precursor.

These results were subsequently correlated with data obtained from absorption spectra taken from each of the individual solutions. Not surprisingly, all of these data imply an interaction between the precursors and the underlying solvent environment. Specifically, UV–visible absorption spectroscopy was utilized to probe interactions between GeI2 and the various solvents, as shown in Figure S6 and summarized in Table S1. In particular, with ethanol, peaks appeared at 291 and 360 nm, whereas with THF, these peaks shifted slightly in position to 294 and 367 nm. With acetonitrile, we observed peaks located at 251, 281, and 328 nm. A single peak situated at 268 nm was found when using a dilute solution of DMF.

These differences in the absorption spectra, coupled with the synthesis data, can be rationalized as follows. Acetonitrile clearly coordinates strongly onto Ge, as our absorption spectra indicate, so it is likely that the solvent inhibits the necessary formation of an intermediate GeI2 layer and that, as a result, MAGeI3 cannot readily crystallize. By analogy, it is evident that, in the presence of an excess of DMF, this solvent strongly coordinates onto Ge when present in excess quantity and as such, essentially completely displaces and replaces the iodide ions. We would expect that spectra similar to that of anhydrous ethanol would be obtained at much higher concentrations of GeI2 in DMF. This assertion is further corroborated by prior work in the literature, describing the synthesis of Ge perovskite films, derived from concentrated 1 M solutions of GeI2 in DMF.[12] Indeed, the spectra from the ethanol and THF solutions were only slightly different, although it should be emphasized that the reaction in ethanol did indeed yield pure MAGeI3. By contrast, with THF, we presume that the iodide was replaced but to a lesser degree as compared with acetonitrile, because it is likely that the GeI2 intermediate and subsequently MAGeI3 both formed. Nevertheless, given our data, it is certainly conceivable that the MAGeI3 emanating from THF likely contained a number of X site vacancies, which would have increased its overall potential to degrade.

Overall, these results indicate that the identity of the good solvent plays a key role in the generation of Ge-based perovskites. Solvents such as DMF and acetonitrile can coordinate strongly onto the GeI2 precursors, thereby inhibiting the formation of constituent GeI6 building blocks, which are crucial for the subsequent production of Ge-based perovskites. In particular, the synthesis of crystals with either residual solvent molecules attached on the surface or a high number of local defects can potentially lead to a fast rate of degradation. Such insights can be used to identify additional solvents and conditions that can be used to generate other types of pure Ge-based perovskites with a relatively low amount of defects and a correspondingly improved stability.

Stability Test

As a means of testing the durability of our as-prepared materials, we conducted a series of stability tests by comparing the XRD patterns of the MAGeI3 and FAGeI3 under different environmental conditions. Specifically, we measured the patterns of samples (i) that were stored in an ambient environment in air versus analogous samples (ii) stored in a glovebox, saturated in an inert argon atmosphere containing <20 ppm oxygen and <0.2% moisture. These samples were both kept under room temperature conditions without either external heating or cooling, a condition which more likely approximates a realistic environment in which the final device would be used. XRD data were collected after exposure times of 1, 3, 6, 24, 48, 72, 100, and 168 h, equivalent in total to a period lasting about a week with the data plotted on a log 10 timescale, so as to achieve a greater appreciation for any time-dependent variations.[35]

Figure A corresponds to the XRD data taken for MAGeI3 stored in a glovebox. It is evident that there was little change up to 6 h and that, even after 24 h, only a slightly broadened peak at ∼20° associated with the (10-2) plane was apparent. The key point is that, overall, the MAGeI3 sample was notably stable with only the appearance of small features at 26.2°, which could likely be ascribed to a GeO2 impurity (indicated by a blue arrow), as confirmed by its standard diffraction pattern.[24] By contrast, the data for the MAGeI3 sample exposed to air were very different (Figure . Specifically, after as little as 24 h of reaction, the diffraction peaks within the XRD pattern associated with MAGeI3 had completely disappeared and were replaced with those of GeO2.

Figure 8

Stability test of as-prepared MAGeI3 and FAGeI3. (A) XRD patterns obtained at different reaction times for MAGeI3 stored inside a glovebox with an inert gas environment; (B) XRD patterns collected at discrete reaction intervals for MAGeI3 stored under air; (C) XRD patterns acquired at different reaction times for FAGeI3 stored inside a glovebox with an inert gas atmosphere; (D) XRD patterns obtained at various reaction times for FAGeI3 stored under air. The timescale shown is presented logarithmically.

A similar scenario played out with FAGeI3. We present the FAGeI3 XRD patterns in Figure C,D for samples stored in glovebox and air environments, respectively. In Figure C, we find that FAGeI3 retained its chemical integrity even after a week since peak positions did not discernibly change; hence, its stability under inert conditions was confirmed. However, corresponding alterations in peak shape and profile reflected a reduction in crystallite size. By contrast, in Figure D, a GeO2 peak located at 26.2° appeared as early as 1 h of exposure. After 24 h, no FAGeI3 was apparent, as it likely had fully decomposed to GeO2. Indeed, both of these germanium perovskites were likely to be very stable when kept under an inert atmosphere, since our samples of MAGeI3 and FAGeI3 retained their native red and yellow colors even after 7 days of storage. By contrast, these identical perovskites were unstable and decomposed in as little as a day in moist air. This net outcome likely included the formation of not only crystalline GeO2 but also amorphous by-products in a process accompanied by perceptible coloration changes to a yellow-brown hue coupled with a structural transformation from a powder to a more gel-like substance.

To mechanistically account for our data, we put forth a plausible explanation for our observations, as follows. The degradation reaction is likely to be a combinatorial effect involving both oxygen and moisture in the air. With the resulting yellow-brown, gel-like amorphous degradation product exhibiting an XRD pattern, which likely indicates the formation of GeO2, we propose a decomposition mechanism, as shown below in reactions and .

Herein, we use MAGeI3 as an illustrative example with which to discuss the degradation reaction pathway. Initially, as per reaction , the as-prepared MAGeI3 perovskite adsorbs H2O onto its surface upon exposure to air; the H atom within the water bonds to the I atom, so as to form a hydrated perovskite structure. Additional water molecules will likely interact with and ultimately displace the MA+ cation, likely through the mediation of the amine group.[25] The remaining Ge complex subsequently decomposes to GeI2 while releasing HI and H2O in the process. This part of the decomposition pathway has been fully explored and is consistent with simulated results, explaining how a MAGeI3 (101) surface can interact with H2O molecules.[26] Nevertheless, the as-produced GeI2 itself is also unstable under an oxygen-rich environment. Reaction suggests that, in the presence of air and oxygen, the iodide can rapidly generate GeO2, which coincidentally is the final product noted in the XRD pattern of Figure B. A similar pathway is expected to explain the decomposition of an analogous FAGeI3 sample with MA+ being replaced with FA+.[5] Overall, our results underline the fact that storage considerations are crucial when engineering devices that incorporate in these Pb-free perovskites. In principle, this problem could be potentially mediated by integrating and incorporating different additive materials, a strategy that has been used in the past to improve upon the stability of both Pb and Sn perovskites.[27]

In order to acquire an improved understanding and more profound insights with respect to the stability of these materials, thermogravimetric analysis (TGA) was employed to investigate the decomposition of these Ge-based perovskites under controlled, high-temperature conditions in different reactive environments. As shown in Figure , samples of MAGeI3 (Figure and FAGeI3 (Figure under a flow of (i) air and (ii) nitrogen gas, respectively, were heated from room temperature to 600 °C.

Figure 9

TGA curves of as-prepared (A) MAGeI3 as well as (B) FAGeI3 under (i) air (black curve) and (ii) nitrogen-based (red curve) atmospheres.

The decomposition pathway under a nitrogen-containing atmosphere has been reported for both MAGeI3 and FAGeI3. In particular, for MAGeI3, the decomposition follows a single-step weight-loss process[28] in which MAGeI3 starts to decompose at 300 °C into MAI and GeI2 with the sublimation process accounted for by the apparent sharp weight loss observed. By contrast, FAGeI3 appears to necessitate two steps for decomposition. The first stage is associated with the production of organic amine molecules, escaping at ∼210 °C,[29] accompanied by a subsequent steep weight loss in a second step, ascribable to the sublimation of GeI2, HI, and methane. In theory, there should be nothing left over. In practice, as can be observed from Figure B, the residue measures ∼4.2%, likely due to either weighting errors or the presence of unknown impurities.

In the presence of air, MAGeI3 still appears to give rise to a single-step weight loss with the decomposition, occurring at a much lower temperature of 120 °C. Specifically, MAGeI3 decomposes into MAI and GeI2, with the GeI2 subsequently oxidized to GeO2. After the heating process, the residue appears to consist of a white powder, determined to be GeO2 via XRD analysis. Indeed, under an air-based atmosphere, the decomposition process happens at lower temperatures, very likely associated with an oxidation process. It should be noted that FAGeI3 gives rise to a lower decomposition temperature (∼100 °C) than MAGeI3. For FAGeI3, the observed weight loss process likely consists of two steps with the first step attributable to the decomposition and loss of both CH(NH)2 and 1/2 I2, contributing to a net 44% weight loss in theory (49%, as experimentally shown in Figure . At 150 °C, the as-produced GeI2 is further oxidized to GeO2 with the concurrent formation of I2 gas, accounting for the observed weight loss. Nevertheless, the residue still consists of GeO2, which is postulated to account for the remaining weight percent of 12% isolated at the end of the process, a value which is very close in magnitude to the experimentally measured 14% quantity, as shown in Figure B.

The results from Figure are comparable with previously reported TGA data for both MAGeI3 and FAGeI3 in a nitrogen-filled atmosphere; we note that no air-based TGA experiments have been reported to date.[30] Overall, our results indicate that our Ge-based perovskites may be less stable than both Sn- and Pb-based perovskites, especially under an air atmosphere.[29,31,32] Nonetheless, because mixed Sn-Ge and Sn-Pb perovskites are well known for evincing better stability, our data may provide significant insights for enhancing the intrinsic stability of Ge-based perovskites.

X-ray Degradation Study (Comparative Analysis of XRD, SEM, TEM, XPS, and SHXM)

As implied earlier, the practical utility of a material is a function to some extent of its inherent stability. Hence, with respect to our as-generated Ge-based perovskites, we chose to analyze this point in a more systematic manner by comparing their behavior when subjected to a broad swath of different irradiation conditions, so as to gain insights into correlations between size, shape, and composition with respect to degradation. The significance of this problem has been addressed to some degree for Pb halide perovskites, whose structural and chemical integrity were impacted by electron-beam- and X-ray-induced decomposition.[33,34] However, as a relatively new material, for Ge-based perovskites, only electron-beam-induced degradation experiments on CsGeI3 have been reported.[13] As such, herein, we have tackled the basic issue associated with how five distinctive, complementary characterization techniques, employing sources of either X-ray or electron beam irradiation, can, in and of themselves, induce perceptible and measurable material transformations within Ge-containing perovskites. In this regard, as one of these methods, we have specifically featured the unique (to the best of our knowledge) use and application of the synchrotron-based scanning hard X-ray microscopy (SHXM) technique to this critical problem.

Indeed, as a means of understanding the durability, stability, and structure of Ge iodide perovskites, synthesized via our facile ligand free coprecipitation method, we employed a relatively new characterization technique, in particular SHXM, with which to study these materials. SHXM affords the opportunity to acquire ultrahigh-resolution X-ray images with a spatial resolution as low as 10 nm, simultaneously coupled with the capability of imaging with chemical sensitivity.[21] Specifically, we conducted fluorescence elemental mapping of MA1–FAGeI3 samples, with x = 0, 0.25, 0.5, 0.75, and 1; the resulting images are shown in Figure . In the far-left column, we present a number of differential phase contrast (DPC) images, highlighting the heterogeneity and nuanced substructure of our samples. In the next two columns, we observe elemental signals due to the Ge K-edge and the I L-edge. These signals clearly spatially overlap with the DPC data. However, the distribution and intensity map of I are dissimilar from those of Ge. In particular, the signal from I appears to be more concentrated within the middle central region of the images, whereas the Ge signal is more diffuse and consequently spread out across the entire structure. This phenomenon is more apparent in the far-right column, wherein RGB images of the Ge K-edge and the I L-edge are presented in red and green, respectively. The pattern of spatial segregation of elements suggests that our Ge perovskites were likely to have degraded under strong X-ray irradiation conditions with Ge2+ reducing to Ge0, coupled with the formation of MAI, FAI, or both. This assertion is corroborated by the detection of Ge0 with our XPS measurements. Furthermore, the Ge perovskite sample shown here appears to be perceptibly larger in size than its nonirradiated counterparts, an observation that can likely be ascribed to the particles melting as a result of intense X-ray beam induced exposure and subsequent degradation.

Figure 10

Nanoscale X-ray images obtained using HXN for MAFA1–GeI3: (A–D) MAGeI3; (E–H) MA0.75FA0.25GeI3; (I–L) MA0.5FA0.5GeI3; (M–P) MA0.25FA0.75GeI3, and (Q–T) FAGeI3. DPC images (A,E,I,M,Q) measure morphology (or electron density) of the sample. Images (B,F,J,N,R) denote the elemental signals associated with the Ge K-edge and (C,G,K,O,S) for I L-edge. Images (D,H,L,P,T) suggest overlap of signals ascribable to Ge K-edge (red) and I L-edge (green). Image (Q) illustrates a central hole feature within the DPC image, with a correspondingly intense Ge signal in panel (R) coupled with a diminished I signal in panel (S).

With respect to the remaining elements, it is plausible that I– may have oxidized into I2 vapor, which was subsequently evacuated from the system, in a process offset by the simultaneous reduction of Ge2+ into Ge0. Indeed, a very similar scenario involving the conversion of Pb2+ into Pb0 accompanied by the complementary oxidation of I– into I2 has been reported for MAPbI3 upon exposure to X-ray beams.[34] Whereas the elemental signatures of the C, H, and N elements associated with the MA+ and FA+ cations are not conclusively detectable using SHXM, our complementary XRD and IR results provide strong evidence for the correct stoichiometry of our as-generated perovskites. Within the DPC image in Figure Q, there is a small hole localized at the center, which was likely induced by X-ray beam damage. To account for this, we note that, within this specified spatial area, there is an apparent and evident Ge signal (i.e., the Ge L-edge) from Figure R overlapping with a correspondingly diminished and weakened I signal (i.e., the I L-edge) from Figure S. These data are consistent with the proposed formation and aggregation of Ge0 coupled with the concomitant oxidation of I– to gaseous I2, ultimately yielding an I-deficient hole.

In terms of obtaining a much closer look at the overall elemental distribution, we conducted a quantitative analysis of the collected data; our results are shown in Figure S8. From this set of processed Ge/I images (with noise reduction), we can clearly visualize the elemental distribution of these two elements, because these types of images purposely emphasize differences or asymmetries in spatial localization. For example, not surprisingly, the Ge/I images will yield a very intense signal if the Ge amount is significantly larger than that of I. Because we observed a relatively low signal intensity within the center of materials, these data implied that the central core possessed a relatively equal quantity of Ge and I. By contrast, the outer shells evinced a much more considerable signal, suggestive of a notably high concentration of Ge as compared with I in that area. As a result, the Ge tends to be more spatially distributed within the outer shell, whereas I is more localized in the midsection. Moreover, the I/Ge images also corroborate the assertion that the presence of Ge is much less dense in the center area, whereas I is more concentrated in the central core region. This prevailing trend, which was observed within all five samples tested, is consistent with the idea that the X-ray beam induced degradation of our Ge-based perovskites leads to a spatial separation and segregation of constituent components, wherein the Ge melts and flows out to the exterior of the material in a process caused by the reduction of Ge2+ to Ge0.

Therefore, our parallel study of CsGeI3 is put forth in Figure . While the DPC image in Figure A implies a structurally heterogeneous sample, the elemental mapping data of the Cs L-edge, the Ge K-edge, and the I L-edge are provided in Figure B–D, respectively. Whereas the Cs and I data overlap very well with each other, the spatial distribution of Ge is clearly different and appears to be more diffuse. The RGB image in Figure E is again consistent with the idea that our Ge perovskites may have decomposed under strong X-ray exposure with Ge2+ reducing to Ge0, concomitant with the formation of MAI, FAI, CsI, or a combination of these species. Similarly, Figure S9 is associated with more quantitative analysis of these images and data. The apparent overlap of Cs and I is consistent with the decomposition of this material into CsI. Furthermore, it should be noted that our Ge/I and I/Ge images yielded similar types of behavior for the set of MA1–FAGeI3 samples. In particular, the Ge component spatially propagated and localized to the outer shell, whereas the Cs and I elements concentrated within the central core.

Figure 11

Nanoscale X-ray images, obtained using HXN, for CsGeI3. (A) Differential contrast phase image. (B) Elemental Cs L-edge; (C) Ge K-edge; (D) I L-edge, and (E) corresponding overlap of all of the three elements associated with Cs L-edge (red), Ge K-edge (green), and I L-edge (blue).

In terms of understanding and rationalizing our collective data, it is worth referring to the prior literature. For instance, with respect to MAPbI3 analyzed with hard X-ray photoelectron spectroscopy (hard-XPS),[34] the Pb 4f region evinced signs that the perovskite phase had decomposed into PbI2 coupled with the X-ray-induced reduction of Pb2+ to Pb0. A similar study of CsPbBr3 subjected to high energy (i.e., 80 to 200 keV) electron irradiation found that the as-prepared CsPbBr3 nanocrystals underwent a radiolysis process in which not only Pb2+ was reduced to Pb0 but also Pb0 subsequently diffused and aggregated.[16] Indeed, these prior literature observations are consistent with our own findings of a Ge0 peak herein, apparent in both the Ge 2p and Ge 3d regions from our XPS data (Figures and ). This X-ray induced decomposition phenomenon is somewhat different from what has been observed in air under ambient storage conditions (i.e., oxidation to GeO2 alone). Indeed, our X-ray-induced degradation noted herein likely proceeds through a stimulated desorption mechanism, wherein I– is oxidized to I0 in the form of I2, thereby leading to the desorption of gaseous I2, in a process accompanied by the reduction of Ge2+ to Ge0.[13,16] The desorption of I0 would account for the lack of any detectable I signal from the XPS measurements, which were collected under ultrahigh vacuum conditions because volatile I2 would not have been present on the surfaces of our perovskites.

Nevertheless, it is worth highlighting that the situation is somewhat more nuanced. Specifically, the quantitative degree of Ge2+ reduction is likely different when using SHXM versus either XPS or XRD, for example, because we did not observe the formation of exactly the same species spanning all three of these characterization methods, which is likely due to a difference in factors, such as the incident X-ray energy, photon flux, focus area, and the flux rate, with the approximate magnitude of each parameter listed for the different X-ray-based techniques tested, as shown in Table S2. A higher flux rate would invariably imply a correspondingly larger degree of X-ray-induced degradation. Not surprisingly, since SHXM combined a high photon flux used (i.e., 109 photons/s) with the smallest focal area probed (i.e., 156 nm2) to yield the largest flux density (i.e., 1024 photons/s·m2, a measure of the number of photons or electrons impinging upon a sample per given unit area over a set period of time) incident upon the sample, this technique gave rise to the highest amount of sample degradation observed.

Furthermore, these results from X-ray-based techniques are also dissimilar from those noted with TEM and HRTEM, which both involve exposure to focused high-energy electrons. These particular data are presented in Figure S7. We found that a combination of both increasing the brightness/intensity and narrowing the focal region of the electron beam could induce not only the reduction of Ge2+ to Ge0 but also the spatial movement and segregation of this element to the central region within the area of irradiation. The phenomenon is consistent with observations from a recent in situ TEM investigation of CsGeI3.[13] By analogy, the observed degree of degradation under these electron microscopy conditions can be potentially rationalized in the context of different irradiation scenarios, as summarized in Table S3. Yet, unlike the X-ray induced degradation, the corresponding electron-beam-induced degradation cannot be explained solely by the dose rate, because TEM and SEM maintain very similar dosing values. An alternative key contributing factor is the acceleration voltage since operating voltages are typically 10 and 80–200 kV for SEM and TEM/HRTEM, respectively. However, relying on this explanation alone may be similarly inadequate. In particular, the situation is likely to be more complex, given that with analogous lead-based perovskites, it has been reported that increased ionization damage was actually observed at the lower acceleration voltage of 80 kV than at 200 kV.[16]

Based on this precedence, one would have expected more damage with SEM than with TEM, but herein, with our Ge-based perovskites, the degree of degradation was higher with TEM as opposed to SEM. As such, to account for our data herein, we hypothesize that, with TEM, the higher acceleration voltages enabled electrons to more readily penetrate through the sample bulk, whereas with SEM, the lower energy electrons mainly impacted the external surface.[36] Therefore, the observed destruction in TEM was greater because the high energy electrons interacted with, permeated, and degraded the entire sample, whereas with SEM, the degree of damage was more spatially confined and therefore limited to the outer surface. In terms of the relevant prior literature, we note that, within a localized beam, associated with even moderate e-beam currents and acceleration voltages, significant degradation for analogous MAPbI3 was observed,[33] again consistent with what we have found herein for our Ge-based perovskites.

Conclusions

We should emphasize that we have not attempted to engineer a series of functional solar cell devices and to optimize their performance. Our work described herein presupposes that insights into the chemistry of these materials are more relevant. In particular, we have proposed and demonstrated the feasibility of a facile, one-step, ambient, ligand-free reprecipitation process for Ge-containing perovskites. This is significant and novel in itself not only because (i) there is a strong motivation to develop sustainable, less-toxic alternatives to Pb-based perovskites for solar applications but also because (ii) most analogous methods, used to synthesize similar types of materials, have tended to be either ligand-mediated or involve multistep, high-temperature operations in inert atmospheres. We have shown that this synthesis protocol is flexible, generalizable, and potentially scalable for ABX3 systems, regardless of the identity of either the A site cation or the X site halide ion. Moreover, our work has emphasized the significance of controlling chemical composition within perovskites as a means of rationally tuning their observed band gaps.

As tangible examples of our efforts in this regard, we have reported herein on the successful production of chemically pure crystalline particulate motifs of not only MAFA1–GeI3, MAGeI3, FAGeI3, and CsGeI3 but also MAGeBrI3– and FAGeBrI3–. We have provided detailed structural information (i.e., microscopy and diffraction) by utilizing TEM, SEM, and HRTEM coupled with XRD for yielding information about crystallinity, morphology, purity, and the nature (i.e., indexing) of exposed lattice planes. Surface properties of our products were analyzed using IR and XPS, so as to confirm their elemental composition, oxidation state, and the nature of their pendant functional groups. Moreover, we performed optical characterization using UV–visible and PL spectral data to highlight that the band gaps of our Ge-containing materials could be systematically varied by altering the ratio between MA+ and FA+ within the A site. Specifically, the photoluminescence profile of our as-prepared Ge-based perovskites yielded emission peaks in the range of 548 to 626 nm with corresponding measured band gaps spanning 2.26 to 1.98 eV, which is well within the favorable absorptive visible region of the solar spectrum. Moreover, to better understand the formation of our materials, we analyzed the effect of good versus poor solvents on the crystallization process. These results indicated that the identity of the good solvent (such as DMF) in particular plays a key role in reliably generating reasonably pure Ge-based perovskites.

To probe the stability of our as-prepared materials, we conducted an XRD-based series of experiments in which the effect of various storage environments was analyzed as a function of exposure time. We deduced that our perovskites are thermally stable under an inert atmosphere but, on the other hand, they may decompose under air. Finally, for the first time with these systems, we were able to compare the effects of electron versus X-ray-induced degradation on our Ge-based perovskites with a particular emphasis on SHXM. Indeed, our SHXM data yielded insights into the evolution of the elemental distribution within the material especially after X-ray exposure. Our data imply that the very act of probing a material may lead to its inadvertent destruction. Indeed, we found a realistic risk of irreversibly degrading Ge-based perovskite species into their constituent components through beam-induced irradiation damage.

Experimental Synthesis Section

Ge-based perovskites have been generated using a facile, room-temperature method based on reprecipitation. Specifically, in this process, germanium (II) iodide (GeI2, Aldrich, ≥99.8%) serves as the Ge source. Conversely, cesium iodide (CsI, Aldrich, 99.999%), methylammonium iodide (MAI, Aldrich, 98%), and formamidinium iodide (FAI, Aldrich, ≥98%) denote precursors incorporating the respective cations, which were investigated. After significant testing, anhydrous ethanol (denatured, HPLC grade, Alfa Aesar, 90%) and n-hexane (spectrophotometric grade, Alfa Aesar, 95+ %) were found to be particularly useful solvents for this family of perovskites.

In a typical reaction, 0.1 mmol of GeI2 and 0.1 mmol of AI (with A = methylammonium, formamidinium, and cesium) precursors were dissolved within 1 mL of anhydrous ethanol so as to form a clear yellow solution. Subsequently, 1 mL of hexane was added as a nonsolvent to precipitate out the desired Ge perovskite. The as-generated variously colored (depending on the perovskite) products (Figure S1) were centrifuged at 9000 rpm for 3 min, collected, and washed with hexane three consecutive times. All of these experimental processes were run under ambient atmosphere and temperature conditions. The final as-prepared sample was stabilized in a hexane solution, placed within a closed container, and finally stored in a glovebox under an Ar atmosphere.

We have systematically analyzed a number of solvent effects. In particular, we replaced the good solvent, that is, anhydrous ethanol, with methanol (CH3OH, Aldrich, 70%), N,N-dimethylformamide (C3H7NO, Sigma-Aldrich, 99.8%), tetrahydrofuran (C4H8O, Sigma-Aldrich, 99.5+ %), acetonitrile (C2H3N, Fisher Scientific, 99.9%), and acetone (C3H6O, Alfa Aesar, 99.5+ %), respectively, in successive experiments. In parallel, we substituted the poor solvent of hexane with toluene (C7H8, Alfa Aesar, 99.5%), carbon tetrachloride (CCl4, Fisher Scientific, 99%), and dichloromethane (CH2Cl2, Sigma-Aldrich, 99.8%) in succeeding runs. The as-generated product was subsequently characterized in terms of both structure and chemical composition using a number of different, distinctive, and complementary techniques, described in detail in the Supporting Information.