Densification-Related Optical and Photodetection Properties of Green-Synthesized MAPbI₃ and MAPbI₃@Graphite Powders

Abstract

For photodetection applications using 3D hybrid perovskites (HPs), dense and thick films or compacted powders in wafer form are needed and generally require large amounts of HPs. HPs are also often combined with a graphene/carbon layer to improve their conductivity. Among HP synthesis methods, mechanosynthesis, a green synthesis method, provides a large amount of powders, which are furthermore easily densified in compact wafers due to their mechanical activation. Thus, methylammonium lead iodide (MAPI) and MAPI with 3 and 5 wt % graphite powders were synthesized, densified by uniaxial compaction, and their photoluminescent (PL) and photodetection properties were studied. MAPI wafers compacted under 100 and 500 MPa showed a PL blue shift compared to the MAPI powder, and photodetection measurements indicated that composite wafers exhibited an enhanced photoresponse with improved photocurrent generation due to the addition of graphite. However, they exhibited weaker photoswitching (on/off) sensitivity with high detected currents in comparison to the MAPI wafer. Such unexpected photodetection behavior with composite wafers were explained by characterizing their microstructural and optical properties. Microstructural characterizations showed no grain coalescence in all MAPI and composite wafers compacted at 100 MPa, but a preferential crystallographic orientation along the {002} plane was detected. Additionally, an unusual graphite segregation near the wafer surface was noticed in composite wafers, questioning the effect of graphite in photodetection performance. Furthermore, all optical measurements evidenced lattice distortions and a decrease of the PL reabsorption phenomenon as a function of the densification pressure, in agreement with the observation of a higher population of photoexcited charge carriers in MAPI compacted at 500 MPa, as observed in transient absorption spectroscopy. These findings demonstrate that green-synthesized MAPI and graphite composites can be easily shaped into dense wafers through simple low-pressure compaction and emphasize the importance of microstructural and optical characterizations to fully understand the resulting properties.

1. Introduction

3D Hybrid perovskites (HPs) have attracted significant research interest in the optoelectronic domain as a key material to fabricate efficient, low-cost, and flexible photonic devices including solar cells, light-emitting diodes, photocatalysts, and photodetectors. , Indeed, 3D HPs are reported to greatly improve responsivity and response speed of photodetectors. , The performance and stability of HP-based photodetectors were reported to be possibly enhanced by incorporating graphite or few-layer graphene. − Such photodetector devices are generally processed in films, the thickness of which may be a key parameter, in particular, for solution-processable methods. Indeed, it is challenging to deposit reproducibly thick polycrystalline films, to control the film morphology, and to ensure a high film density. Additionally, solvents used in these methods may introduce harmful traces of water, form intermediate phases, and induce nonstoichiometry that may influence the stability of HPs and performance reproducibility in devices. There are thus always challenges in the processing/shaping of HPs for such detection applications, and furthermore, strategies involving powders also deserve to be studied and developed. Some key steps in such powder-based strategies are the production of large amounts of powders and their processing/shaping.

In that context, mechanosynthesis (MS), involving the high-energy milling of the reactants, enabled production of gram-scale powders at room temperature in a reproducible way and is thus promising to address these challenges. It is also a solvent-free (green) synthesis that can be carried out under an inert atmosphere and has already been demonstrated as a HP promising synthesis method. − Therefore, MS can be considered an appealing method for the synthesis of methylammonium lead iodide (MAPI) powders for such applications. In addition, MS powders have been reported to lead to higher green density (the density measured after the simple compaction step) in comparison with unground powders due to their refined grain size but also to their ≪more reactive surfaces≫ improving their compressibility. − Generally, compacted powders are often submitted to heat treatment to improve their density and crystallinity. However, heat treatment can be complicated with HPs and especially MAPI, which undergo phase changes at a low temperature (around 55 °C). Furthermore, such uniaxial compaction of HP powders in wafers may affect their optical and microstructural properties as pressure effects have been reported during in situ measurements under high pressure on single crystals. − To the best of our knowledge, there are only a few studies reporting only the simple uniaxial compaction effect of such green-synthesized HP powders on their structural and optoelectronic properties. Witt et al. − studied the compaction dynamics of MS MAPI by conducting time-dependent pressure measurements, illustrating the relative density evolution as a function of pressure and time. A high relative density of 95% was reached by compacting powders under 100 MPa for 360 min. In another similar study on the effect of the starting powder grain size on the compaction process, Witt et al. pointed out a general decrease in the crystallite size of MAPI when the pressure increases and an increased PL lifetime in larger grain wafers accompanied by a PL peak at higher energy attributed to reduced defects in larger grains. Witt et al. , also studied the grain size improvement under pressure and high temperature: they reached a relative density higher than 97% with an average grain size of 1.9 μm under 100 MPa, 100 °C and observed an enhanced preferred orientation along the (002) direction with the increase of temperature. Zheng et al. reported an electrical and mechanical field-assisted sintering technique (FAST) of ball-milled MAPI powders involving a pressure of 52 MPa and a high temperature of 200–300 °C and reported an increase in grain sizes, partial stabilization of a cubic $\mathit{Pm}\stackrel{-}{3}m$ phase, and better photodetection performance.

Thus, few studies were carried out on the effect on microstructural and optical properties of a simple uniaxial compaction of MAPI powders. − Furthermore, such studies could trigger the development of powder-based strategies to fabricate some HP-based devices. Recently, we have established the MS conditions of MAPI powder and MAPI composite powder with 5 wt % of graphite, which have been shown promising for photodetection. Indeed, this dry process provides an intimate mixing of components leading to efficient interfaces between HP and graphite, promising for photodetection applications. We have at first tested different graphite concentrations up to 26 wt %. However, the photoluminescence (PL) spectra showed significant PL quenching at a high graphite content (Figure S1) due to the strong light-absorbing property of graphite. Thus, when the graphite amount is higher than 5 wt %, the graphite strongly affects the optoelectronic properties of MAPI. That is why 3 and 5 wt % were selected for this study. Concerning the selected pressures, in situ high-pressure measurements on single crystals − reported phase transformation from 300 MPa; as earlier powder compaction studies dealt with pressures of several hundred MPa and considering the good densification of mechanosynthesized powders at low pressures, the two pressure values have been chosen below (100 MPa) and above (500 MPa) 300 MPa.

Therefore, we have examined here the effect of compaction on the structural and optical properties of MAPI powders densified under two pressures (100 and 500 MPa) and on MAPI composite powders with 3 and 5 wt % graphite densified under 100 MPa and then evaluated their photodetection properties. All powders and wafers were characterized structurally by X-ray diffraction (XRD) and scanning electron microscopy (SEM), and their optical properties were characterized by UV–vis absorption spectroscopy, static and time-resolved PL, and transient absorption (TA) spectroscopy. PL measurements showed a blue shift of PL peaks after the compaction step of MAPI increasing with the applied pressure. Photodetection measurements confirmed the relatively better photoresponse under irradiation in composite wafers due to the presence of graphite but showed a weaker on–off photosensitivity in composite wafers in contrast with the MAPI wafer. The combination of photodetection measurements, microstructural and optical studies, as well as surface-sensitive X-ray photoelectron spectroscopy (XPS) measurements on both powders and wafers allowed evidencing a graphite segregation at the wafer surface and a reduced reabsorption phenomenon in the PL spectra induced by the compaction step and suggesting that graphite mixed with HPs may affect the on–off photosensitivity for HP photodetector applications. Such results put in view the potential of MS to synthesize reproducibly and provide large amounts of HPs and composite powders and to obtain dense wafers by compaction at low pressure, thus enabling easily investigating different powder-based applications of HP.

2. Experimental Section

2.1. MS of MAPI

The MS conditions of MAPI are reported in our previous studies. In brief, the precursor powders and graphite were weighted according to the stoichiometry and then transferred into a stainless-steel milling jar filled with stainless-steel balls in a glovebox under an argon atmosphere. The ball-to-powder weight ratio R remained at 40. The ball milling was also carried out under argon conditions by using a laboratory planetary micromill (Fritsch PREMIUM PULVERISETTE 7). The grinding speed was set at 350 rpm for 30 min. The obtained powders were named as MAPI30.

2.2. MS of MAPI@Graphite Composites

Graphite powder Timcal (Imerys group), MAI, and PbI₂ (stoichiometric ratio MAI:PbI₂ = 1:1) powders were introduced in the vial under an argon atmosphere. The mixture was ground at 350 rpm for 30 min with a ball-to-powder weight ratio of 40. The amount of graphite was varied in the reactant mixture. 79.83 mg of graphite, 0.662 g of MAI, and 1.919 g of PbI₂ were mixed together for the preparation of MAPI composite with 3 wt % graphite (named MAPI@G3). 0.133 g of graphite, 0.648 g of MAI, and 1.880 g of PbI₂ were mixed together for the preparation of the MAPI composite with 5 wt % graphite (named MAPI@G5). All syntheses were realized under an argon atmosphere, and the resulting powders were stored in a glovebox under an argon atmosphere.

2.3. Wafer Fabrication

Each standard powder specimen (0.10–0.11 g) was precisely weighed and compressed under 100 MPa pressure by using a uniaxial press with a 7 mm diameter cylindrical mold made of steel. The wafer was under pressure for 3 min, and the wafer shaping was realized under ambient air condition. MAPI30 samples were also pressed under 500 MPa for the study of the compaction pressure effect. After compaction, all samples were stored in the glovebox. The wafers are cylindrical and displayed a diameter of 7 mm and a thickness of around 0.7–0.8 mm.

2.4. XRD

XRD patterns were collected using a BRUKER D8 DISCOVER diffractometer equipped with a copper X-ray anode (Kα1 = 0.154056 nm), a motorized antiscatter screen, and a monochromator. The measurements employed a θ–θ Bragg–Brentano geometry and a Lynxeye XE-T energy-resolved linear detector to filter the fluorescence. For MAPI and its composites, the diffraction patterns were measured in the 5 to 90 deg range in 2θ with a step of 0.02 degree. Same amounts of samples were used for each XRD measurements. The nature of crystalline phases was identified by XRD. Lattice parameters and mean crystallite size were obtained from experimental data using Fullprof software. Le Bail analysis was performed, and Thomson–Cox–Hastings function was chosen for the profile. An internal standard instrumental contribution was measured by a corundum.

2.5. UV–Visible Diffuse Reflectance Spectroscopy and Band Gap Determination

Ultraviolet–visible diffuse reflectance spectroscopy was performed by using a PerkinElmer Lambda 950 UV/vis spectrometer equipped with an integration sphere. The spectral range was from 300 to 900 nm. The measured reflectance spectra were transformed into corresponding absorption spectra by applying the Kubelka–Munk function: $F(R_{\infty })=\frac{K}{S}=\frac{(1-R_{\infty }{)}^2}{2R_{\infty }}$ where K stands for the absorption coefficient and S is a semiempiric scattering coefficient. R _∞ is the reflectance of an infinitely thick specimen: $R_{\infty }=\frac{R_{\mathrm{sample}}}{R_{\mathrm{standard}}}$ . The band gap energy determination is based on the Tauc method where energy-dependent absorption coefficient α can be expressed by (α · hν)^1/γ = B(hν – E _g), where h is the Planck constant, ν is the photon’s frequency, E _g is the band gap energy, and B is a constant. The γ factor equals to $\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.$ because MAPI is a direct-band semiconductor. By replacing α with F(R _∞) and plotting (F(R _∞) · hν)^1/γ as a function of hν, the x-axis intersection point of the linear fit of the Tauc plot gives an estimated band gap energy.

2.6. PL Measurement

The measurements were performed with a 320 nm wavelength excitation source using a Horiba Fluorolog spectrofluorometer. The powders or wafers were fixed between two quartz windows. Small quantities of powder (around 10–20 mg) were loaded for each measurement.

2.7. SEM

SEM was used to observe as-mechanosynthesized powders and compacted powders using a Zeiss Gemini SEM 500 scanning electronic microscope operating at 3.00 kV. The wafers were broken manually by rupture to access the cross-section.

2.8. Time-Resolved PL

Samples were photoexcited at 2 kHz using 400 nm excitation, and epifluorescence PL was collected with a 100 mm focal length lens. These photons were then directed to a 1/8 m spectrograph and a single-photon sensitive streak camera.

2.9. TA Spectroscopy

TA was performed using a 35 fs, 800 nm, 2 mJ/pulse, amplified titanium:sapphire laser operating at 2 kHz. Probe pulses were produced using a portion of the laser output that was mechanically time-delayed and then focused into a 2 mm thick sapphire crystal. Probe pulse light reflected off the sample at a ∼ 10° angle of incidence was then dispersed and measured with a high-speed array detector. Pump pulses were reduced in repetition rate to 1 kHz with a mechanical chopper and then frequency doubled in a BBO crystal and attenuated with a variable neutral density filter.

2.10. Photodection

Photodetection tests were carried out on MAPI30, MAPI@G3, and MAPI@G5 wafers compacted under 100 MPa by directly connecting them to two tungsten tips from the setup. The I–V characteristics of the samples were obtained using an AAA rated ORIEL Verasol-2 LED solar simulator (LSS-7120) in AM1.5G conditions and a Keithley 2461 source measure unit. 100 mW/cm² irradiance corresponds to 1 sun. The photoresponsivity R _λ at 5 V bias was calculated by applying equation $R_{\lambda }=\frac{I_{\mathrm{light}}-I_{\mathrm{dark}}}{P_{h\nu }{A}_{\mathrm{s}}}$ , where R _λ is photoresponsivity, I _light is the current, I _dark is the dark current, P _hν is the input photon power (100 mW/cm²), and A _s is the effective surface around 7 mm diameter at wafer surface. The detectivity (D*) was determined by using the equation $D^{*}=\frac{R_{\lambda }}{\sqrt{2q\frac{I_{\mathrm{dark}}}{A_{\mathrm{s}}}}}$ , where q is the elementary charge. The photoswitching on–off tests were realized at 5 V bias under AM1.5G conditions by manually switching on and off the solar simulator every 30 s.

2.11. XPS

XPS experiments were conducted in an ultrahigh-vacuum spectrometer equipped with a RESOLVE 120 MCD5 hemispherical electron analyzer. The XPS spectra presented here were obtained using Al Kα radiation of a twin anode source and pass energies of 100 and 20 eV for the survey and high-resolution spectra, respectively. All XPS spectra were calibrated to 284.8 eV according to the C 1s binding energy for adventitious carbon or adsorbed surface hydrocarbon species from the atmosphere. Analysis of the spectra was done with Casa-XPS software with the application of Shirley background subtraction and a combination of Gaussian–Lorentzian symmetric and asymmetric line shapes for the peak fitting.

3. Result and Discussion

3.1. Effect of the Densification Step on MAPI30 and Composite Powders

MAPI powders obtained by high-energy ball milling were named MAPI30 and compacted under uniaxial pressures of 100 and 500 MPa (named MAPI-100 MPa and MAPI-500 MPa, respectively). MAPI composite powders with 3 and 5 wt % of graphite were obtained through the same standard 30 min one-pot MS, and their compacted wafers under 100 MPa were named MAPI@G3-100 MPa and MAPI@G5-100 MPa, respectively.

3.1.1. Microstructure of MAPI30 and Composite Wafers Compacted at 100 and 500 MPa

The MS MAPI30 powders display a fractal microstructure: they are in the form of micrometric agglomerates of aggregates of grains (400 ± 152 nm), themselves constituted of oriented nanograins (nanograin size ≈ 5–10 nm) (Figure S2 and Table ). , After the compaction step, all MAPI30 wafer samples display a density of around 3.7 to 4 g/cm³, which is around 90 to 95% of the one of MAPI single crystal (around 4.15 g/cm³, estimated from lattice parameters). Such high relative density was also reported by Shrestha et al. (90%) and Witt et al. (95%) in quite stronger pressure conditions. Both MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers display also a very densely compacted feature with a density around 3.8 to 4 g/cm^–3.

1. Lattice Parameters, Lattice Volumes, Crystallite Size, and SEM Mean Grain Size of MAPI30 Powder, MAPI-100 MPa, MAPI-500 MPa, MAPI@G3-100 MPa, and MAPI@G5-100 MPa.

	lattice parameter (±0.001 Å)
sample	a = b	c	lattice volume (Å3)	XRD mean crystallite size (±1 nm)	SEM mean grain size (nm)
MAPI30 powder	8.874	12.669	997.7	189	400 ± 152
MAPI-100 MPa	8.874	12.672	997.9	434	388 ± 135
MAPI-500 MPa	8.861	12.643	992.7	too weak peak broadening	521 ± 187
MAPI@G3-100 MPa	8.872	12.669	997.2	117	106 ± 55
MAPI@G5-100 MPa	8.873	12.669	997.4	114	109 ± 69

^aThe crystallite size of MAPI-500 MPa could not be extracted from the XRD pattern because the full width at half-maximum (fwhm) of the XRD peaks was too small to be analyzed by Le Bail method as it was close to the instrumental fwhm.

The microstructure of MAPI30 wafers compacted under different pressures (100 and 500 MPa) was compared through SEM imaging. The wafers were broken manually to observe the internal microstructure of the cross-sections. The SEM images of MAPI-100 MPa and MAPI-500 MPa wafer cross-sections in Figures , and S2A,C confirm the dense microstructure of both wafers with almost no porosity. SEM images collected from the backscattered electron (BSE) mode confirm a good compositional homogeneity all along the section (Figure S3B,D). The microstructure of the different MAPI@G wafers with graphite doping amount of 3 and 5 wt % is displayed in Figure . Graphite flakes can be clearly recognized in the BSE mode SEM images of wafer cross-section as they appear in black color, while MAPI is brighter (Figures C,D and S4). The graphite sheets are homogeneously distributed along the central cross-section. However, one notices that the density of graphite flakes appears higher at the near-surface cross-section of both MAPI@G3-100 MPa and MAPI@G5-100 MPa (Figure E,F, highlighted by rectangles). This evidences a partial segregation of graphite under applied pressure, leading to a higher graphite amount at the wafer surface. To further confirm the graphite segregation at the wafer surface, XPS was also performed as a surface-sensitive technique to probe the surface composition. According to the survey scan (Figure S5), an atomic ratio of 10.5 between carbon and lead, C/Pb, was found in MAPI@G5 as-ground powder. A strong increase of the C/Pb atomic ratio was noticed on both sides of the MAPI@G5-100 MPa wafer: from 10.5 (MAPI@G5 powder) to 26.8 and 38.0 (face 1 and face 2 of MAPI@G5-100 MPa, respectively). These C/Pb ratios indicate a higher amount of graphite on the MAPI@G5 wafer surface compared with the raw powder, which is in good agreement with the observed graphite segregation by SEM.

1.

SEM images of wafer “central” cross-sections of (A) MAPI-100 MPa and (B) MAPI-500 MPa and SEM images of wafer near-surface cross-sections of (C) MAPI-100 MPa and (D) MAPI-500 MPa (dotted rectangles highlight the grain coalescence region).

2.

SEM images of wafer central cross-sections of (A) MAPI@G3-100 MPa and (B) MAPI@G5-100 MPa, BSE mode images of wafer central cross-sections of (C) MAPI@G3-100 MPa and (D) MAPI@G5-100 MPa, and BSE mode images of wafer near-surface cross-sections of (E) MAPI@G3-100 MPa and (F) MAPI@G5-100 MPa (dotted rectangles highlight the graphite segregation region).

An average grain size of 388 ± 135 nm is deduced from the cross-sectional SEM images of MAPI-100 MPa wafers (Figures A, S3A, and Table ) to compare with the mean grain size of 400 ± 152 nm in MAPI30 powders (Figure S2 and Table ). Both mean grain sizes are quite close, suggesting that the grain size has not been influenced by this compaction step under 100 MPa pressure. In contrast to MAPI-100 MPa, the MAPI-500 MPa wafers exhibit a higher mean grain size of 521 ± 187 nm (Figures B, S3B, and Table ), which suggests that coalescence between grains has taken place at this higher pressure. This grain coalescence is visible in the SEM image of the central cross-section of the MAPI-500 MPa wafer with a distinct grain size larger than that of MAPI-100 MPa (Figure A,B). Moreover, as detailed in the Supporting Information (SI) part, the microstructure of the wafer cross-section close to the mold surface is also analyzed (Figures C,D and S6) because during the uniaxial compression process, the pressure is generally not homogeneous in the whole compacted powder volume (the pressure decreased with depth) and is higher at the wafer surface close to the “piston.” While no grain coalescence is observed close to the MAPI-100 MPa wafer surface (Figure C), some grain coalescence is noticed under 500 MPa pressure as revealed in Figure D and SI part.

Compared with the SEM grain size of MAPI-100 MPa, a strong decrease in the grain size was observed in both MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers (Figure A,B and Table ). As observed with MAPI30 powder densified under 100 MPa, the mean grain size of MAPI@Graphite powders (Figure S7) around 101 ± 36 nm is similar to that in wafers (106 ± 55 and 109 ± 69 nm for MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers, respectively). However, the mean grain size is around 3.5 times lower for MAPI@Graphite wafers than for the MAPI-100 MPa wafer (Table ). We attribute it to the addition of graphite, a well-known lubricant, that should reduce the welding process during ball milling and enhance the fracture mechanism. In conclusion here, as for MAPI30, the compaction step under 100 MPa has not affected the grain size of MAPI@Graphite, and no grain coalescence has been observed.

3.1.2. Crystallographic Structure of MAPI30 Wafers Compacted at 100 and 500 MPa

The XRD patterns of MAPI30 powder and wafers and MAPI@Graphite composite powders and wafers in Figure display the characteristic XRD peaks of the tetragonal I4/mcm MAPI phase. However, an additional small peak at around 26° corresponding to the diffraction of graphite is observed in both MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers (Figure B) but not in the XRD pattern of the powders. This XRD peak at 26° is the basal reflection (002) of graphite (corresponding to a d-spacing of 0.335 nm). Meanwhile, the XRD detection limit is about 5 wt %, which indicates that as we did not identify the graphite XRD peak in XRD pattern of MAPI@Graphite composite powders, we should not be able to detect the presence of graphite if it was homogeneously distributed in the wafer volume. Thus, observing the graphite diffraction peak in wafer samples indicates a higher graphite concentration on the wafer surface. This result agrees well with the observed graphite segregation in the SEM images at the near-surface cross-section of both wafers and with the XPS results showing a higher amount of graphite in MAPI@G5-100 MPa wafer surface than in the powders.

3.

XRD pattern of (A) mechanosynthesized MAPI30 powder and MAPI-100 MPa and MAPI-500 MPa wafers and (B) mechanosynthesized MAPI@G3 powder, MAPI@G3-100 MPa wafer, MAPI@G5 powder, and MAPI@G5-100 MPa wafer.

For the MAPI-100 MPa, MAPI@G3-100 MPa, and MAPI@G5-100 MPa wafers, the relative intensity of (002) [compared with the (110) peak intensity] and (004) [compared with the (220) peak intensity] XRD peaks at around 14.0° and 28.2°, respectively, increases, supporting a crystallographic preferential orientation along the plane family {002} (Figure ). The evolution of intensities ratio between (002) and (110) XRD peaks is detailed in the SI part and supports a crystallographic preferential orientation along the plane family {002} for all wafers. This preferential orientation is more pronounced in the MAPI-500 MPa wafer (Figure ), as illustrated by the enhanced relative intensity of (002) and (004) XRD peaks at around 14° and 28.2°, respectively. Indeed, the intensity ratio between (002) and (110) XRD peaks increases from 0.76 up to 3.06 and that between (004) and (220) XRD peaks also increases and more strongly from 0.72 up to 5.94 for MAPI-100 MPa and MAPI-500 MPa, respectively.

4.

Zoom-in XRD pattern of MS MAPI30 powder and MAPI-100 MPa and MAPI-500 MPa wafers.

In addition, for MAPI-500 MPa, the XRD peaks located at around 28.2°, 28.5°, 31.6°, 31.8°, and 31.9° corresponding to planes (004), (220), (114), (222), and (310) become asymmetric after wafer shaping, and an additional diffraction peak not belonging to the tetragonal I4/mcm structure at 14.05° was noticed. These XRD peak shoulders and the additional peak suggest the appearance of a different crystallographic structure in the MAPI-500 MPa wafer. Phase transitions in MAPI single crystals are reported by in situ high-pressure X-ray diffraction techniques which demonstrated the appearance of a cubic phase at a high pressure. ,, However, our XRD peaks do not match this high-pressure cubic phase. Instead, they rather match with a high-temperature cubic $\mathit{Pm}\stackrel{-}{3}m$ phase observed by Zheng et al. within their MS MAPI wafers developed through a FAST process. Different from the high-pressure cubic phase, the $\mathit{Pm}\stackrel{-}{3}m$ scubic phase is highly ordered without octahedral distortion. In our case, the asymmetry of (004), (220), (114), (222), and (310) XRD peaks in the MAPI-500 MPa wafer (Figure ) could be assigned to a slight amount of this metastable cubic phase as its characteristic (200) and (210) XRD peaks are located between (004) and (220) XRD peaks and between (114) and (222) XRD peaks of the room-temperature tetragonal phase, respectively. Especially the extra peak at 14.05° in the MAPI-500 MPa wafer, located between (002) and (110) XRD peaks of the tetragonal phase, may be attributed to the (002) XRD peak of this cubic phase. Therefore, this XRD analysis implies the appearance of a $\mathit{Pm}\stackrel{-}{3}m$ cubic phase in the MAPI30 wafer compacted under 500 MPa.

The lattice parameter and crystallite size of the tetragonal I4/mcm phase were also deduced from the analysis of the XRD patterns of MAPI and composite wafers (Figure S8). They all show similar lattice parameters and lattice volumes (Table ), indicating that the crystallographic structure of MAPI30 would not be affected by the compaction step. Nevertheless, the lattice volume of the MAPI-500 MPa wafer is slightly smaller (992.69 ų instead of 997.89 ų) suggesting a small lattice shrinkage under 500 MPa compaction. The crystallite size increases from 189 ± 1 nm to 434 ± 1 nm after compaction of MAPI30 powder under 100 MPa. Considering the standard deviation on SEM mean sizes, the MAPI-100 MPa crystallite size is rather close to the SEM mean grain size, 388 ± 135 nm, suggesting that the grains in the MAPI-100 MPa wafer are monocrystalline, which is not the case for MAPI30 powder grains which are rather polycrystalline (lower crystallite size than SEM mean grain size). Thus, a coalescence between the polycrystalline domains within MAPI30 grains is suggested to take place resulting in the increase of crystallite size leading to monocrystalline grains. With MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers, similar results are obtained with their crystallite sizes being similar to their SEM grain sizes, confirming that the grains are monocrystalline in wafers.

Thus, the compaction at 100 MPa does not modify the grain size but should favor the oriented aggregation of nanograins in grains, leading to grain monocrystallinity. Under 500 MPa, the smaller lattice volume suggests a stronger pressure effect on the MAPI lattice under 500 MPa compaction. The crystallite size of MAPI-500 MPa is also bigger (Table ). This proves again a grain coalescence extending to crystallite size induced by high pressure. Moreover, by comparing the two first intense (002) and (110) XRD peaks of MAPI-100 MPa and MAPI-500 MPa wafers at around 14° and 14.15°, a peak narrowing is noticed (Figure ) in MAPI-500 MPa confirming the higher crystallite size in MAPI-500 MPa. The obtained XRD results are thus in good agreement with the increase of grain size observed in the SEM images.

3.1.3. Optical Properties of MAPI30 and Composite Wafers Compacted at 100 and 500 MPa

Figure A presents the PL spectra of MAPI30 powder and wafers compacted at 100 and 500 MPa. A clear shift of the PL peak position toward higher energy is observed when the powders underwent a compaction step. The main emission PL peak position of MAPI30 powder is located at 1.59 eV, while the ones of MAPI-100 MPa and MAPI-500 MPa are shifted to 1.62 eV. However, the PL peak of MAPI-100 MPa is very broad and exhibits a shoulder toward lower energy. It has thus been deconvoluted into two peaks: the main peak at 1.62 eV and one side peak at 1.52 eV (Figure S9B). Similarly, a side peak at 1.52 eV may also be identified in MAPI-500 MPa but with a much weaker intensity (Figure S9C), and a PL side peak in MAPI30 powder may also be pointed to at 1.62 eV (Figure S9A).

5.

PL spectra of (A) MS MAPI30 powder and MAPI-100 MPa and MAPI-500 MPa wafers and (B) MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers.

The PL spectra of MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers are noisy (Figures B and S10). This is in agreement with the already reported PL quenching due to the presence of graphite and the resulting charge transfer at graphite-MAPI interfaces. The main emission peak of MAPI@G3-100 MPa wafer is at 1.62 eV but one broad side peak at 1.51 eV is noticed (Figure S9D). MAPI@G5-100 MPa displays a similar emission behavior with a main peak at 1.63 eV and a shoulder peak at 1.51 eV (Figure S9E). The composite powders exhibit similar PL spectra in Figure S10, but the PL peak at 1.50 eV displays a higher intensity than in the wafers. Thus, a decrease in the intensity of the PL peak at about 1.50 eV is also observed with compacted composite powders as with MAPI30 wafers. Nevertheless, one may notice that the PL spectra of MAPI and composite powders are different: MAPI powder displays a main peak at 1.59 eV (with a shoulder at 1.62), while composite powders display a main peak at ≈1.62 eV with a very broad shoulder peak at ≈1.51 eV. Therefore, for MAPI, a blue shift of the PL peak appears to occur after compaction when, for the composites, as soon as the pressure increases, the intensity of the PL peak at ≈1.51 eV decreases. All these results suggest a pressure-driven phenomenon.

To elucidate the “blue shift” of PL peaks or the decrease of the PL peak intensities at low energy after the densification step of MAPI and composites, respectively, further UV–vis analyses are carried out (Figures S11–S14). At first, the UV–vis absorption spectra of composite wafers are very noisy, as observed with PL spectra, due to the presence of graphite (Figure S13). Absorption of MAPI can be roughly remarked between 1.5 and 1.6 eV (Figure S13), but the band gap E _g values in MAPI@G3-100 MPa and MAPI@5-100 MPa wafers cannot be deduced from Tauc plots (Figure S14). Furthermore, the Tauc plots of MAPI@G3-100 MPa and MAPI@5-100 MPa wafers are very close to the one of graphite (Figure S14A). Meanwhile, the band gap of MAPI@G3 and MAPI@G5 powders can still be analyzed (Figure S14B), giving slightly higher band gap values than the one of MAPI30 powder (1.54 eV): 1.57 and 1.58 eV, respectively. They are still in the reported MAPI band gap value range (1.5–1.6 eV). The nondetectable MAPI band gap in composite wafers supports again the higher concentration of graphite on the wafer surface, and the strong absorbing characteristic of graphite screened the signal of MAPI.

For MAPI30, the E _g value obtained from the Tauc plots decreases from 1.54 to 1.45 eV for MAPI30 powder and both wafers, respectively. Such a decrease of band gap is often reported during experiments of MAPI single crystal under pressure and may thus be explained by some deformations in MAPI30 grains upon compaction. One may effectively expect disorder of the crystal lattice caused by variations in bond length and bond angle, which agrees well with the lower lattice volume observed with the MAPI-500 MPa wafer. The comparison of PL and UV–vis absorption spectra of MAPI30 wafers evidence a clear anti-Stokes shift (Figure S11), which is not observed with MAPI30 powders (Figure S12B). All these results suggest that the band gap decrease and the PL peak evolution (blue shift) after densification are not correlated and that the band gap decrease is certainly due to lattice distortion as often reported. The higher band gap with composite powders by comparison with MAPI30 powder could be related to less strong mechanical chocks during the mechanosynthesis due to the presence of graphite acting as a lubricant.

To further investigate the influence of compaction in MS MAPI30 powder, the charge recombination dynamics under illumination is studied through time-resolved photoluminescence (trPL) for MAPI30 powder and MAPI-100 MPa and MAPI-500 MPa samples. The trPL spectra in Figure (under 400 nm excitation wavelength, from the beginning of the recombination process up to 800 ps) demonstrates that a much faster charge recombination process occurs with wafers in comparison with MAPI30 powder. The three decay curves are fitted with a double-exponential model:

$$y=y_0+A_1\exp (-x/{\tau }_1)+A_2\exp (-x/{\tau }_2)$$

where y ₀ is an offset (here y ₀ is fixed to 0), A ₁ and A ₂ are weight constants, and τ₁ and τ₂ are time decay constants. The fitting results of decay times and the corresponding amplitudes are listed in Table . MAPI-100 MPa and MAPI-500 MPa wafers display similar lifetimes, and their lifetimes τ₁ and τ₂ are lower than those of MAPI30 powder. The attribution of these lifetimes to radiative or nonradiative recombination related to defects differs a little bit from one report to another. The shorter lifetimes τ₁ and τ₂ of HPs deduced from a fitting with a double-exponential model are assigned to interfacial charge transfer and trap-assisted recombination, respectively, or to free carrier recombination due to surface effect and in the bulk, respectively. The smaller lifetime for compacted powders suggests that the compaction step has affected the recombination rate, notably the trapping process, and reduced the carrier lifetime of MS MAPI30 even at moderate pressures (100 and 500 MPa). Nevertheless, the lifetimes for the compacted wafers are quite similar. This may be correlated with disorder or lattice distortion as deduced from UV–vis measurements, which increases after the compaction step but is similar for both applied pressures. As the band gap is also shown to be similar for both pressures, one may suggest that the pressure has induced similar lattice distortion.

6.

trPL dynamics of band edge emission from MS MAPI30 powder, MAPI-100 MPa wafer, and MAPI-500 MPa wafer.

2. Charge-Carrier Lifetimes and Amplitude of Dynamics Extracted from trPL Spectra from 0 to 800 ps of MAPI30 Powder, MAPI-100 MPa Wafer and MAPI-500 MPa Wafer.

samples	MAPI30 powder	MAPI-100 MPa	MAPI-500 MPa
A 1	1.0 ± 0.1	2.4 ± 0.1	2.2 ± 0.1
τ1 (ps)	78 ± 5	44 ± 2	46 ± 2
A 2	0.5 ± 0.1	0.4 ± 0.1	0.4 ± 0.1
τ2 (ps)	974 ± 61	588 ± 13	548 ± 13

trPL at a longer time scale was also performed (Figure S15). Surprisingly, the trPL decay curve remains very similar on both powder and wafer-shaped samples, suggesting that the used pressure in this study (100 and 500 MPa) mainly changes the fast component during the charge-carrier recombination, notably the trapping process, while the influence on slower trPL recombination is found to be very weak. This agrees with the amplitude A ₂ values which are smaller than the A ₁ values. However, the trPL can be affected by several different recombination processes inside the material. These can include the recombination between the photoexcited charges and the intrinsic-doped charge carriers, which may have resulted from compaction-induced deformations and disorder.

Therefore, to better characterize the photogenerated charge carriers, TA spectroscopy was carried out on both MAPI-100 MPa and MAPI-500 MPa. A femtosecond-to-nanosecond time window was measured in order to evaluate the fast and slower carrier dynamics. TA measurements could not be performed on the powder sample due to their high roughness and low reflective nature. TA spectra under 3.1 eV (400 nm) excitation from 0.5 to 20 ps of both compacted samples in Figure show similar characteristics of reported spectra for MAPI. , Two photobleaching (PB) bands were found at around 1.62 eV (765 nm) and 2.58 eV (480 nm), named PB1 and PB2, corresponding to the state filling. Between PB1 and PB2, a broad band of photoinduced absorption was observed. The peak of the sharp PB1 band representing the band filling between the minimum of conduction band and the maximum of valence band was observed to shift to the lower energy when the time increases, indicating photoinduced band gap narrowing on both compacted wafers. The comparison of TA spectra of MAPI-100 MPa and MAPI-500 MPa at different time delays in Figure S16 shows no obvious difference in the shape and position of the PB1 band between two different pressures.

7.

TA of (A) MAPI-100 MPa wafer and (B) MAPI-500 MPa wafer from a 0.5 to 20 ps time delay.

To further investigate the TA process of the band gap state filling, the kinetics at 1.6 eV (773 nm), corresponding to the peak position at 5 ps time delay of both MAPI-100 MPa and MAPI-500 MPa, from 0 to 3000 ps was evaluated. Figure shows the beginning of the TA dynamics, which is a zoom-in part of Figure S15 (TA dynamic curve of 0–3000 ps). By testing double- and three-exponential models (Figure S17), the three-exponential model turned to fit better the decay curves for both samples:

$$y=y_0+A_1\exp (-x/{\tau }_1)+A_2\exp (-x/{\tau }_2)+A_3\exp (-x/{\tau }_3)$$

where y ₀ is an offset (here y ₀ is fixed to 0), A ₁, A ₂, and A ₃ are weight constants, and τ₁, τ₂, and τ₃ are time decay constants. This fitting with a three-exponential model suggests that three processes are happening during the photobleaching. The obtained lifetime and amplitude from the fitting of the TA dynamic curve of a window time of 0–3000 ps are summarized in Table . The lifetime τ₁ remains similar between two wafers, the one of MAPI30-500 MPa is slightly 1.3 ps longer than that of MAPI-100 MPa, while τ₂ of MAPI-500 MPa (164.3 ± 21.3 ps) is much longer than that of MAPI-100 MPa (94.2 ± 10.0 ps). The slowest lifetime τ₃ of MAPI under lower pressure (100 MPa) also shows close to the one under 500 MPa if considering the fitting uncertainty (3371.2 ± 215.5 and 2964.5 ± 244.8 ps for MAPI-100 MPa and MAPI-500 MPa, respectively).

8.

Dynamics extracted from TA of MAPI-100 MPa wafer and MAPI-500 MPa wafer at 773 nm (1.60 eV).

3. Charge-Carrier Lifetime and Amplitude of Dynamics Extracted from TA Spectra of MAPI-100 MPa Wafer and MAPI-500 MPa Wafer at 773 nm (1.60 eV).

sample	MAPI-100 MPa	MAPI-500 MPa
A 1	0.8 ± 0.1	0.6 ± 0.1
τ1 (ps)	8.5 ± 0.3	9.8 ± 0.5
A 2	0.2 ± 0.1	0.19 ± 0.02
τ2 (ps)	94.2 ± 10.0	164.3 ± 21.3
A 3	0.2 ± 0.1	0.3 ± 0.1
τ3 (ps)	3371.2 ± 215.4	2964.5 ± 244.8

Now looking at the amplitude of the fitting results, for the fast process, the amplitude of MAPI-100 MPa (0.8 ± 0.1) is higher than the one of MAPI-500 MPa (0.6 ± 0.1), and the second medium rate process amplitude stayed similar for two pressures. As the amplitude of the TA decay spectra signifies the photogenerated charge population, the obtained results demonstrate that a higher amount of charges was generated in the MAPI-100 MPa wafer during the first fast relaxation channel and the charge population is close between two different pressures for the second process. However, a higher amplitude level A ₃ was found with MAPI-500 MPa for the longest process with a value of 0.3 ± 0.1 in comparison to the A ₃ = 0.2 ± 0.1 of MAPI-100 MPa, which could also be clearly noticed on the decay curve. This implies that MAPI-500 MPa might be able to photogenerate more long-lifetime charge carriers under the same condition in comparison to MAPI-100 MPa.

3.2. Photodetection Measurements on Mechanosynthesized MAPI and MAPI@Graphite Composite Wafers Compacted under 100 MPa

I–V curve measurements under dark and different simulated solar irradiation intensities (Figure S18A) between −5 and 5 V show that the dark current of MAPI-100 MPa varies between 10^–12 A and 10^–9 A. As already observed in our previous work, a minimum current shift is always observed for MAPI-100 MPa (in particular for dark and low irradiation intensity measurements), and a slight photocurrent is noticed in I–V curves between −5 and 5 V (Figure S18): the current at 5 V bias under 100 mW/cm² (1 sun illumination) is around ≈2.6 × 10^–9 A for MAPI-100 MPa. A near linear dependency of current on the illumination intensity is remarked (Figure S18B) indicating a photoresponse of the MAPI-100 MPa wafer. The photoresponsivity and detectivity values (cf. equation in experimental part) of MAPI-100 MPa and MAPI@G wafers are compared in Table . MAPI-100 MPa wafer displays quite a low photoresponsivity (4.4 × 10^–8 A/W) in agreement with its low photocurrent certainly related to its low electrical conductivity. Similarly, the obtained detectivity D* of MAPI-100 MPa remains relatively low compared to those of MAPI@G wafers with a value of 1.6 × 10⁶ Jones, showing thus a lower detection sensitivity.

4. Photoresponsivity of MAPI-100 MPa, MAPI@G3-100 MPa, and MAPI@G5-100 MPa at 5 V Bias under 100 mW/cm² Illumination.

sample	MAPI-100 MPa	MAPI@G3-100 MPa	MAPI@G5-100 MPa
R λ (A/W)	4.4 × 10–8	2.8 × 10–5	1.4 × 10–2
D* (Jones)	1.6 × 106	2.7 × 107	1.2 × 109

For both MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers (Figure ), the minimal current is centered at 0 V in contrast to MAPI-100 MPa. The noise in I–V curves is reduced with a current increase certainly due to the addition of graphite, and a continuous increase of photocurrent with the voltage bias is also observed with composite wafers (Figure ). A photocurrent is noticed for MAPI@G3-100 MPa as a function of illumination power: at 5 V bias, 1.3 × 10^–6 A under dark and 2.4 × 10^–6 A under 100 mW/cm² illumination (Figure ). For MAPI@G5-100 MPa, we obtained higher values at 5 V bias, 1.7 × 10^–4 A under dark and 7.0 × 10^–4 A under 100 mW/cm² illumination, certainly due to its higher amount of graphite on comparison with MAPI-100 MPa and MAPI@G3-100 MPa. These results show that 5 wt % graphite provides better conductivity and charge transfer than 3 wt % graphite. Moreover, the photocurrent response at 5 V bias of MAPI@G5-100 MPa composite wafers increases with the irradiation power (Figure C). Meanwhile, for MAPI@G3-100 MPa at 5 V bias, an increase of photocurrent is noticed when the irradiation intensity is lower than 40 mW/cm² and a photocurrent saturation is observed at higher irradiation conditions (Figure C). Such a photocurrent saturation is an often-observed phenomenon of photodetectors, possibly related to the filling of sensitizing centers (localized states that can capture one type of photogenerated charges) at high illumination intensity. , These results demonstrate better and more effective photoconduction in composite wafers and especially in MAPI@5G-100 MPa. This is also confirmed by their much higher R _λ values compared with MAPI-100 MPa in Table . The higher R _λ of MAPI@G5-100 MPa (1.4 × 10^–2 A/W) implies a better photoconductivity potential compared with MAPI@G3-100 MPa (2.8 × 10^–5A/W). Furthermore, MAPI@G5-100 MPa displays the highest D* value (1.2 × 10⁹ Jones), which supports an improved photodetection sensitivity when MAPI is efficiently combined with 5 wt % of graphite through a one-pot MS.

9.

I–V curves under the dark and different irradiances of (A) MAPI@G3-100 MPa and (B) MAPI@G5-100 MPa wafers. (C) Photocurrent at 5 V bias of MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers under dark and different irradiances (inset: zoom of I–V curve of MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers from −3 to −5 V).

The sensitivity of the sample on photoswitching (on and off) under 100 mW/cm² was also tested on all wafers. An on–off photosensitivity is observed with MAPI-100 MPa (Figure S19A), thus confirming the effective photoresponse of the MAPI-100 MPa wafer under 100 mW/cm² irradiation (Figure S18B). By contrast, the on–off photoswitching ability of MAPI@G3-100 MPa and MAPI@G5-100 MPa appears affected/perturbated as shown in Figure S19B. For the MAPI@G5-100 MPa wafer, the increase of current when the light is “on” at around 10s and the decrease of current at the light-off state at around 60s can be observed, but with the on–off curve, it is very noisy. Similarly, the MAPI@G5 on–off curve is also very noisy. The current increase at around 20, 80, and 135 s when the light is switched “on” is remarked, while the decrease of current without illumination is less distinct. The detectable response to “on–off” irradiation with MAPI@G3 and MAPI@G5 wafers agrees with the current dependency on the illumination intensity observed in their I–V curve (Figure ). Their less sensitive and weaker response to photoswitching compared with the MAPI-100 MPa wafer is surprising and may thus be linked to the addition of graphite and certainly the above observed graphite surface segregation in wafers. Furthermore, one notices also very high values of detected current during the photoswitching measurements increasing with the graphite amount (Figure S19B). This less-resolved on–off curves could be related to these high current values inducing possibly some “heating phenomena” perturbing thus the photoswitching measurements.

In terms of the stability of the samples, our previous study demonstrated that no degradation of MAPI@G5-100 MPa wafer was noticed by XRD after 7 weeks of storage in a nonsealed container. Furthermore, the I–V photoresponse remained similar to that of “fresh” sample, showing quite a good long-term stability of the MAPI@G5 composite. In addition, to check the sample state after light exposure, dark currents of MAPI-100 MPa and MAPI@G5-100 MPa were measured immediately once again after 1 to 100 mW/cm² exposure. The dark current of MAPI-100 MPa after light exposure is very close to the initial dark current (Figure S19A), confirming the stability of the MAPI-100 MPa wafer. Meanwhile, for the MAPI@G5 wafer, the dark current became slightly higher in the low-voltage domain (in the range −2 to 2 V) after light exposure (Figure S20B), implying that the combination of higher current and irradiation has modified the material, and such a phenomenon is often referred to as light soaking.

Photodetection measurements (I–V curves under different irradiancies) show that both MAPI and composite wafers display a photoresponse under irradiation and that the photoresponsivity increases when graphite was added with a quite good long-term stability. However, the “on–off” photosensitivity is observed to be strongly affected in composite wafers on comparison with the MAPI wafer, and very high current values are also noticed. Considering the graphite segregation at the wafer surface confirmed by SEM, XRD, and UV spectroscopy, one may suggest that this surface graphite accumulation after compaction would be responsible for the low “on–off” sensitivity of composite wafers.

3.3. Discussion on the Compaction Effect on MAPI30, MAPI@G3, and MAPI@G5 Powders

Microstructural characterizations have shown that the compaction step under 100 MPa did not modify the SEM grain size of MAPI30 and composites but led to monocrystalline grains (instead of polycrystalline grains as observed in MAPI30 powders) and their preferential orientation along the {002} planes. The grain and crystallite sizes were smaller in composite powders and wafers on comparison with MAPI ones due to the lubricating effect of graphite. The increase in the compaction pressure at 500 MPa of MAPI30 powder enhanced the preferential orientation, induced both grain and crystallite growth, and led to the appearance of a new cubic phase already identified by Zhang et al. Interestingly, the observed cubic $\mathit{Pm}\stackrel{-}{3}m$ phase, that does not belong to the conventional high-pressure phase appeared only for the compaction at 500 MPa.

The grain and crystallite size variation under compaction is not easy to rationalize due to the fact that few studies were performed just on a simple compaction step, and most of them combined a high temperature or dealt with the crystallization of MAPI film assisted by pressure and/or temperature. Shrestha et al. observed a similar crystallite size as the one of their starting microcrystal MAPI powders after 300 MPa compaction. Witt et al. observed the decrease of MAPI crystallite size when the wafers were compacted only under 100 MPa with multiple compression and decompression cycles. They attributed the crystallite size reduction to grain plastic deformation during the compaction. However, most of the time, a size increase was reported. ,,− Therefore, one may conclude that the simple uniaxial compaction step of ground powders would induce grain growth only under pressures above 100 MPa and at least of 500 MPa.

The preferred orientation of grains in wafers is less discussed in published studies under such moderate-compaction condition. Except few studies demonstrating preferential orientation along (110) and (220) planes for the film obtained under hot isostatic pressing or made by drip-pressing, most of the reported results showed a preferred orientation along (002) and (004) planes after simple pressing or hot pressing, ,, which is in good agreement with our results with all wafers. Moreover, our results demonstrated that by increasing the pressure from 100 to 500 MPa, the (002) and (004) crystallographic plane orientations are further enhanced. One may suggest that the arrangement of grains during the compaction step would trigger a preferential orientation.

Nevertheless, a graphite segregation was observed by SEM in the BSE mode, close to the wafer surface in MAPI@G3-100 MPa and MAPI@G5-100 MPa wafer’s cross-sections. It was further confirmed by the higher C/Pb atomic ratio on the wafer surface than in the powder, as deduced from XPS measurements, by the appearance of the XRD peak of graphite in XRD patterns of wafers and not of powders, by noisy PL spectra due to charge transfer between graphite and MAPI, and by UV–vis absorption spectra of wafers exhibiting mainly the characteristics of graphite. As graphite was homogeneously distributed within powders, this result suggests a “migration” of graphite sheets toward the wafer surface. To further confirm the “migration” of graphite toward the surface, a compaction of a simple mixture of MS MAPI and 5 wt % graphite without ball milling step was carried out. The entire cross-section of the wafer is given in Figure S21. A distinct separation of graphite from MAPI grains was noticed (highlighted by rectangles in Figure S21). Almost all of the graphite sheets were located at the wafer near-surface part, leaving MAPI powder in the wafer center. This observation evidences that under pressure, the graphite flakes would migrate toward the wafer surface. Compared with the strong segregated wafer made from a simple mixture, fewer graphite sheets are observed at the surface of composite wafer showing a better mixing and stronger interface interactions between MAPI and graphite in the composites synthesized by ball milling. Graphite being a well-known lubricant, one may expect that the graphite sheets, in the powder aggregates near the surface during the compaction stage, slipped and thus migrated toward the surface under the effect of the pressure gradient in the compacted volume.

Due to the graphite segregation in composite wafers, the optical characterizations have been centered on MAPI30 wafers. PL measurements have shown that after compaction of MAPI30 powders, a blue shift of the main PL peak is observed, which increases with the increase of compaction pressure. The composite powders display a main PL peak at around ≈1.62–1.63 eV with a shoulder at 1.51 eV, and after compaction, the intensity of the low energy PL peak is observed to decrease for both wafers with 3 and 5 wt % of graphite. The comparison of PL spectra with the UV–vis absorption spectra of MAPI30, MAPI-100 MPa, and MAPI-500 MPa, showed an absorption red shift (Figure S12A). The band gap values decrease after compaction of MAPI30 powders and are similar whatever the pressure. It is commonly reported in pressure-dependent optical in situ studies on single crystals that before 300 MPa, there is a band gap narrowing with the increase of pressure due to the decrease of Pb–I band length, and after the phase transition, the band gap jumps to a higher value and continues to increase at higher pressures. Similarly, the PL peak position varies in the same way as the band gap with a first red shift and then a blue shift at a very high pressure. − By contrast, Bonomi et al. studied the ambient retention of pressure-induced effects on MAPI by using a press allowing a very fast pressure release, and they observed for both E _g and PL peaks a blue shift after pressure release increasing as a function of the applied pressure. They attributed this blue shift to quenching of the high-pressure state. Zhang et al. reported an increase of the band gap and also such a blue shift of the PL peak after compacting and releasing pressure on MAPI nanoplates and ascribed it to the long-range order damage of the inorganic framework of MAPI under pressure. Zheng et al. (FAST process) attributed the PL blue shift to reduced tail states near the band edge. From all these results, we suggest that the anti-Stokes shift observed from the comparison of UV–vis and PL spectra is mainly due to a narrowing of the band gap and resulted from the applied pressure during the compaction step, inducing stresses and plastic deformations that lead to lattice distortion such as the decrease of the Pb–I bond length. However, the reported evolution of the PL spectra under pressure and/or combined with other treatments did not allow us to explain our PL results and suggests that the observed PL “blue shift” could not be related to direct pressure effects similar to those on band gap evolution. Indeed, both UV–vis and PL spectra would have evolved similarly if they both resulted from a pressure effect.

From our recent study, we considered the involvement of a reabsorption phenomenon that induces the PL “red shift” or the presence of low-energy PL peak. Indeed, we recently demonstrated, by performing PL measurements on MS powders as a function of particle size ranges as well as by surface characterizations, that the ball-milling process induces surface modification of MAPI grains leading to a reabsorption phenomenon. The intensity of this reabsorption PL peak was observed to increase when the particle size decreases. The presence of a surface defective layer was confirmed by combining different surface characterization techniques including XPS, EDX, positron annihilation spectroscopy, TA spectroscopy, and static and trPL. Smallest particles exhibited enhanced reabsorption due to stronger surface effects. Furthermore, such a reabsorption phenomenon was observed in MAPbBr₃ single crystal and in MAPI films ,− with the observation of red-shifted PL peaks. As the PL peak position of MAPI without reabsorption would be located at around 1.62–1.63 eV from theoretical calculations, , the main PL peak of MAPI30 powder at 1.59 eV would mainly be due to reabsorption as the reabsorption phenomenon leads to emission at lower energy and the weak shoulder PL peak at 1.62 eV would be attributed to the PL emission peak of MAPI. This suggests thus that a strong reabsorption phenomenon would occur in the MAPI30 powder. After compaction, the PL peak is shifted to higher energy, suggesting a strong decrease of the reabsorption phenomenon. The compaction step seems to have “annihilated” the surface effect, leading to the decrease of reabsorption as well as a higher contribution of the MAPI PL.

Similarly, with MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers, we observed a similar decrease after compaction in the intensity of the PL peak attributed to this reabsorption phenomenon, and this confirms again a decrease of the reabsorption phenomenon when powders are compacted. This is in agreement with the results of Diab et al. on MAPbBr₃ single crystals and Witt et al. on pressed MAPI films, who also reported on double-emission peaks in HPs: the higher-energy one was attributed to the surface emission, while the lower-energy one was attributed to reabsorption. More precisely, Witt et al. associated the main peak at 1.61 eV to PL that came directly from the surface sample and the shoulder at about 1.53 eV to a reabsorption effect. Therefore, we attribute such reduced reabsorption with both MAPI and composite compacted wafer to a decrease of the surface effect.

Thereby, we can conclude that the “blue shift” observed in MAPI-100 MPa and MAPI-500 MPa wafers is actually not a real emission blue shift but is due to a reduced reabsorption phenomenon and enhanced MAPI PL emission induced by the compaction step. Such a decrease of reabsorption under densification is a quite interesting result because a reabsorption-related PL change was frequently reported in hybrid perovskite film or single crystals and the enhanced PL peak red shift was more related to a higher thickness. −

trPL spectra were fitted by a biexponential law, and both τ₁ and τ₂ lifetimes decrease after the compaction step and are quite similar, whatever the applied pressure. However, lifetimes are often reported to increase after the compaction or hot-pressing step due to mainly the grain coalescence. ,,, One may thus notice that the grain and crystallite size increase induced by the highest compaction pressure has not improved the lifetimes, as reported in published studies. This different behavior in our case might be explained by a higher level of distortion/disorder or defects introduced by the grinding step, which should further increase after the compaction step, in agreement with the observed band gap narrowing. During the compaction, grains underwent plastic deformations , leading certainly to higher concentrations of distortions/disorder inducing more nonradiative charge recombination. Furthermore, TA spectroscopy measurements performed on MAPI wafers showed an increase in the photogenerated charge carrier amount in MAPI-500 MPa in comparison with MAPI-100 MPa. This result agrees with the disappearance of the reabsorption phenomenon when the pressure increased, especially at 500 MPa. Considering the microstructural evolution during the compaction step, the disappearance of the PL reabsorption peak and the increase in the photogenerated charge-carrier amount in MAPI30 wafers when the pressure increases may be related to the obtained dense microstructure and/or to the observed preferential orientation in wafers, which increased with higher compaction pressure limiting thus the reabsorption phenomenon.

Finally, the composite powders were shown to exhibit higher band gap values by comparison with MAPI powders and two distinct PL peaks in PL spectra, even after compaction, with the more intense peak always being the high-energy peak (which was not the case for MAPI30 displaying one main peak at low energy). These results suggest that the reabsorption and distortion phenomena were less important in the composite powders. We attribute these results to the presence of graphite, acting as a lubricant and limiting the impact of mechanical shocks during the synthesis process and compaction.

Photodetection measurements have shown that the presence of graphite was beneficial to improve electron transport in wafers but was detrimental to the photoswitching behavior. This graphite segregation should explain the absence of on–off photoswitching response of composite wafers on comparison with the MAPI30 wafer (inducing possibly a surface graphite percolation and a local heating), but we will also have to check if such low photoswitching response could be solely due to the presence of graphite.

4. Conclusions

In this work, we have performed the uniaxial compaction of green-synthesized MAPI and composite MAPI@graphite (3 and 5 wt %) powders and then structural and optical characterizations and photodetection measurements to study the densification effect in such a powder-based wafer fabrication approach.

The microstructural and optical characterizations and compositional analysis by XPS performed on wafers have evidenced graphite segregation near the wafer surface. Furthermore, except for the appearance of a preferential (00l) grain orientation in all wafers, the grain size was not affected by the compaction step under 100 MPa. By contrast, grain coalescence and cubic phase were observed under 500 MPa.

The optical characterizations and analyses have shown that the band gap narrowing observed after compaction was mainly due to lattice distortion induced by the applied pressure. The PL measurements have evidenced a blue shift of the PL peak after the compaction step of MAPI powders, when for composite powders, two PL peaks were identified with the intensity of the PL peak at low energy decreasing after compaction. The analysis of PL results allowed the conclusion that the “blue shift” observed after MAPI compaction was not a real emission blue shift but was due to a reduced reabsorption phenomenon and enhanced MAPI PL emission triggered by the compaction step. Such a decrease of reabsorption after compaction also explained the decrease of the low-energy PL peak intensity after composite powder compaction. It explained furthermore the higher population of photoexcited charge carriers in MAPI compacted under 500 MPa as evidenced by TA spectroscopy. This decrease of the reabsorption may be related to the densification of powder grains and/or to the preferential grain orientation, but certainly to an annihilation of surface effects under pressure, and further structural and defect studies are needed to conclude. Finally, the lower effect of compaction on optical properties of composite powders (lower narrowing of band gap and lower reabsorption) was related to the presence of graphite, which should preserve the powders from strong deformations and surface defects.

Finally, photodetection measurements have shown that the MAPI-100 MPa wafer presented a photoresponse under illumination and on–off photoswitching behavior. By contrast, MAPI@Graphite wafers displayed improved photoconductivity and photoresponsivity explained by the addition of graphite. However, these composite wafers exhibited less-resolved photoswitching behavior in contrast to the MAPI wafer and furthermore very high current values, in particular, for MAPI@G5-100 MPa. The graphite segregation may explain such a low photoswitching behavior, but a question arises if the presence of graphite itself could be responsible for such a lack of an on–off signal. Moreover, the electrode configuration (tungsten tip directly pointing at the bare wafer surface) could also be improved. Further experiments could be carried out by depositing silver electrodes at the wafer surface and carrying photodetection measurements by further reducing the amount of graphite. Surface characterizations such as XPS could also be performed to further analyze the graphite segregation.

These results highlight the interest of shaping hybrid perovskite powders to develop new HP-based devices but also emphasize the importance of studying the effect of powder shaping on optical and microstructural properties. Finally, this work demonstrated that as a green and effective synthesis method, MS could provide activated powders that are easy to densify at low pressure and exhibit enhanced surface photoactivity.

Glossary

Abbreviations

HP

hybrid perovskites

MAPI

methylammonium lead iodide

MS

mechanosynthesis

PL

photoluminescence

XRD

X-ray diffraction

UV

ultraviolet

Vis

visible

SEM

scanning electronic microscopy

trPL

time-resolved photoluminescence

TA

transient absorption

XPS

X-ray photoelectron spectroscopy

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c02836.

• PL spectra of MAPI@Graphite composite wafers with different graphite concentrations; SEM image of MAPI30 as-ground powder (insert: grain size distribution curve); lower-magnification SEM images of wafer cross-section of MAPI-100 MPa and MAPI-500 MPa and the corresponding BSE images of MAPI-100 MPa and MAPI-500 MPa (insert: grain size distribution curve); BSE mode SEM images at higher magnification of wafer central cross-sections of MAPI@G3-100 MPa and MAPI@G5-100 MPa; SEM images of wafer near-surface cross-sections of MAPI-100 MPa and MAPI-500 MPa; SEM image of MAPI@G 3.5 wt % powder (insert: size distribution curve); Le Bail refinement of mechanosynthesized MAPI, MAPI@G3, MAPI@G5 powders, MAPI-100 MPa, MAPI-500 MPa, MAPI@G3-100 MPa, and MAPI@G5-100 MPa wafers; PL curve deconvolution of MAPI30 powder, MAPI-100 MPa wafer, MAPI-500 MPa wafer, MAPI@G3-100 MPa wafer, and MAPI@G5-100 MPa wafer; PL spectra of MAPI@G3 powder and MAPI@G3-100 MPa wafer, MAPI@G5 powder and MAPI@G5-100 MPa wafer; UV–vis absorption spectra of mechanosynthesized MAPI30 powder and MAPI-100 MPa and MAPI-500 MPa wafers and UV–vis absorption and PL spectra of MAPI30 powder; UV–vis absorption spectra of MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers; Tauc plot of MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers and graphite powder and MAPI@G3 and MAPI@G5 powders; trPL spectra over a 50 ns time window for mechanosynthesized MAPI30 powders, MAPI-100 MPa wafer, and MAPI-500 MPa wafer; TA spectra of MAPI30-100 MPa wafer and MAPI30-500 MPa wafer at different time delays; triple-and double-exponential fitting of dynamics extracted from the TA spectra of MAPI30-100 MPa wafer and MAPI30-500 MPa wafer at 773 nm (1.60 eV); I–V curves under the dark and different irradiances of mechanosynthesized MAPI30 and photocurrent at −2 V bias of MAPI30 wafer under dark and different irradiances; photoswitching curves of MAPI-100 MPa and MAPI@G3-100 MPa and MAPI@G5-100 MPa wafers at 5 V bias under 100 mW/cm2 illumination; I–V curves under the dark of mechanosynthesized MAPI30 and MAPI@G5 wafers before and after irradiation; and SEM image of a wafer made of MAPI and 5 wt % graphite simple mixture (PDF)

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Y.C.: Writing–original draft, investigation, methodology, formal analysis, and data curation. A.A.A.P.: Investigation. D.B.: Funding acquisition and methodology. C.L. and S.K.: Methodology. T.F.: Methodology, data curation, and validation. R.D.S.: Methodology, investigation, and validation. C.S.: Conceptualization, methodology, and validation. S.B.-C.: Conceptualization, methodology, funding acquisition, project administration, and validation

We thank the French Ministry of Research (fellowship to Yihui Cai) and the Fondation Jean-Marie Lehn (SBE-003-23) for the financial support. Work performed at the Center for Nanoscale Materials, a U.S. Department of Energy Office of Science User Facility, was supported by the U.S. DOE, Office of Basic Energy Sciences, under contract no. DE-AC02-06CH11357.

The authors declare no competing financial interest.