Shaping perovskite materials into micro/nanostructures for photonics and optoelectronic applications

Abstract

Metal halide perovskites have emerged as a highly versatile class of materials for optoelectronic and photonic applications, combining exceptional photophysical properties with the ability to accommodate variations in composition, crystal structure, and dimensionality. They can be synthesized under mild solution conditions as either polycrystalline thin films or high-quality single crystals, the latter offering superior structural purity and reduced defect densities that are ideal for advanced device applications. This perovskite structural adaptability, together with compatibility with diverse fabrication strategies, enables the creation of complex micro- and nanoscale architectures with finely tunable optical and electronic properties. This review examines recent advances in growth and morphological control approaches aimed at producing high-quality perovskite structures with tunable geometries. In particular, it highlights how deliberate shaping and structural control of metal halide perovskite materials can unlock new functionalities and device architectures, positioning perovskites as a highly adaptable platform for next-generation optoelectronic and photonic technologies.

Engineering the shape of metal halide perovskite thin films and single crystals through advanced growth and fabrication strategies enables tailored optical properties and innovative optoelectronic and photonic device architectures.

Introduction

Metal halide perovskites (MHPs) have recently emerged as one of the most fascinating materials for optoelectronic and photonic applications. The perovskite structure is characterized by a three-dimensional crystal lattice composed of corner-sharing BX₆ octahedra, with the general formula ABX₃.^1,2 In this structure, A represents a monovalent cation that can be a metal ion or an organic molecule such as Cs⁺, methylammonium (MA⁺), formamidinium (FA⁺), or a larger organic cation, such as phenethylammonium (PEA⁺) or butylammonium (BA⁺). The B site is occupied by a divalent metal cation, typically Pb²⁺ or Sn²⁺, and X denotes a halide anion (i.e. Br⁻, I⁻ or Cl⁻). These components determine the dimensionality and symmetry of MHP materials, which can be categorized as three-dimensional (3D), two-dimensional (2D), one-dimensional (1D) or zero-dimensional (0D),^3,4 each with its own distinct structural, optical, and optoelectronic properties.^5–7

In contrast to conventional inorganic semiconductors, MHPs combine superior optoelectronic performance with exceptional flexibility that allows variations in composition, crystal structure, and dimensionality.⁸ Relevantly, MHPs can be prepared through a variety of deposition techniques, such as solution processing,^9,10 or solvent-free approaches like thermal evaporation and chemical vapor deposition (CVD).¹¹ The possibility to be deposited via different techniques enables the formation of either polycrystalline films or high-quality single crystals.^12,13 The latter offer superior structural purity and low defect densities, making them particularly well suited for advanced optoelectronic and photonic applications.^14,15 This inherent structural adaptability, combined with the compatibility of MHPs with a broad range of fabrication approaches, enables the creation of geometrical shapes and complex architectures with finely tunable optical and optoelectronic properties.¹⁶

To this scope, in recent years, a variety of top–down and bottom–up micro- and nanostructuring techniques have been established to produce micro- and nanoscale arrays in both MHP single crystals and polycrystalline films.

(i) Top–down methods are post-synthesis fabrication techniques, such as photo- and electron-beam lithography (EBL),^17–19 which can in principle provide precise geometries at the micro- and nanoscale. However, these techniques are usually combined with the use of polar solvents or etching chemistry, that may lead to significant material degradation and require special processing conditions. For this reason, alternatives to conventional lithographic methods, such as micro- and nanoimprint lithography (MIL/NIL) top–down fabrication approaches, have recently been proposed to define micro- and nanoscale MHP structures, providing simple, efficient, and non-destructive methods.²⁰

(ii) Bottom–up methods form the desired micro- and nanostructures in situ during MHP crystallization and generally preserve crystal quality more effectively. Geometrically confined crystallization method, for example, relies on polymeric molds or silicon templates to shape the precursor solutions into predefined microstructures.²¹ This offers straightforward geometrical control and has recently enabled the fabrication of subwavelength structures with dimensions in the nanometer range.²² Another bottom–up route takes advantage of patterned substrates with hydrophilic sites to geometrically localize the growth of MHP structure, without the use of any template.²³ Although this technique has demonstrated functional device integration, the resulting geometries often have irregular shapes.

In this scenario, the review provides a comprehensive overview of the unique properties of MHP materials and highlights recent advances in morphological control and growth strategies targeting high-quality, tunable structures for applications ranging from optoelectronics to photonics. This review will also highlight how the deliberate shaping and micro- and nanostructuring of MHP materials can unlock new opportunities in device architecture, further expanding their potential as a highly versatile platform for next-generation optoelectronic and photonic technologies. By enabling geometric control, spatial confinement, and tailored micro- and nanoscale organization, structured MHPs can exhibit functionalities that are difficult or in some cases impossible to realize with conventional semiconductor systems, thereby driving innovation in both material design and device performance.

Strategies for the fabrication of precisely structured MHPs

Post-growth approaches

This section discusses post-growth, top–down micro- and nanofabrication techniques used to structure MHP materials. Many approaches can be used to obtain the desired structural and functional properties. The choice of method depends on several factors, including the target device application, the required feature size and resolution, the composition and stability of the MHPs, and the fabrication throughput. Careful consideration of these parameters is essential to optimize the performance and reliability of MHP-based devices. The following discussion reviews the main fabrication strategies, outlining their advantages, limitations, and suitability for different classes of MHP materials and device architectures. In general, three main classes of post-growth top–down techniques can be identified for structuring MHP materials: (i) mask-mediated indirect approaches, (ii) direct milling approaches, and (iii) micro- and nanoimprint lithography. Mask-mediated indirect approaches include lithographic methods that use an intermediate mask or template on pre-grown MHP layers to define patterns. A subsequent chemical or physical process, such as reactive ion etching (RIE), transfers the pattern into the MHP layer. These methods are well established and provide high precision and control over feature size and geometry. However, they often require multiple fabrication steps and may expose MHPs to damaging conditions, including solvents, high temperatures, reactive species, or high-energy electron beams. Direct milling techniques, in contrast, create nanostructures directly on pre-grown MHP films or single crystals without resist masks or additional processing. They typically rely on focused beams (e.g., ions or lasers), mechanical probes, or localized energy sources to remove or modify material with nanoscale precision. While these methods simplify processing and reduce contamination risks, they may introduce surface damage, ion implantation, or scalability limitations. Micro- and nanoimprint lithography offer an alternative strategy for MHP patterning. These techniques reduce surface damage and enable reliable pattern transfer with resolutions beyond conventional photolithography, while remaining compatible with large-area and high-throughput fabrication. Together, these approaches represent the main strategies for MHP micro- and nanostructuring, each with specific advantages and trade-offs in terms of precision, material compatibility, and process integration.

Mask-mediated indirect approaches

Electron beam lithography (EBL) is one of the most widely used techniques for nanostructuring materials with nanometer-scale features. Standard EBL processing involves the deposition of electron-beam-sensitive resists followed by a soft-bake step to remove residual solvents, which is resist-dependent but is typically carried out at temperatures well above 100 °C. After exposure, polar solvents are employed during the development and lift-off steps. The combined action of elevated temperatures and polar solvents can seriously damage MHP materials, whose extreme sensitivity requires the development of dedicated lithographic approaches, significantly deviating from standard processes.

Despite these limitations, EBL can still be exploited in top–down micro- and nanostructuring, relying on direct etching of MHPs performed through a resist-based physical mask. Microdisk lasers and waveguide-coupled photodetectors, shown in Fig. 1A, were manufactured in such a way, exploiting standard 20 KV EBL processing on poly(methyl methacrylate) (PMMA) resist, baked at 60 °C, which is lower than the standard temperature, for an hour, followed by inductively coupled plasma (ICP) etching of single crystal MAPbBr₃ plates.¹⁹ The etching was performed with a mixture of chlorine gas and C₄F₈. The former plays the role of the main etchant specie, while carbon ions protect the structure sidewalls during etching, and fluorine ions react with the redundant chloride ions, preventing underetching. Although the development of PMMA in this case is based on methyl isobutyl ketone (MIBK): 2-propanol (IPA), which is known to affect MHPs, the authors report an extremely fast development time (5 seconds) and highly reproducible device performance. A similar recipe was also used to nanostructure MAPbX₃ (X = Br⁻, I⁻ or Cl⁻) plates into coupled waveguide-microdisk architectures with separation down to 140 nm (Fig. 1B)²⁴ and 1D gratings²⁵ using ZEP 520A as the e-beam resist and n-amyl acetate as the developer.

Fig. 1. (A) MAPbBr₃ single-crystal microdisk and grating-coupled waveguide patterned by EBL on PMMA followed by ICP etching. Adapted with permission from ref. 19 Copyright 2017, Wiley-VCH. (B) MAPbBr₃ microdisk coupled to a bus waveguide (140 nm gap) fabricated by EBL and ICP etching on a microplate. Adapted with permission from ref. 19 Copyright 2017, Wiley-VCH. (C) Orthogonal photolithography patterning of MHPs using a negative orthogonal photoresist and SF₆ RIE. Adapted with permission from ref. 26 Copyright 2016, Wiley-VCH. (D) Patterned MHP structures based on CsPbBr₃ (green), MAPbI₃ (red), and CsPbCl₃ (blue) obtained using a PMMA sacrificial layer and positive photoresist, with pattern transfer by RIE. Adapted with permission from ref. 17 Copyright 2025, American Chemical Society. (E) Dual-color photolithography using PMMA as a protective layer and SU8 photoresist, showing patterns on the same substrate plane (left) and on an inverted SU8 film (right). Adapted with permission from ref. 27 Copyright 2019, American Chemical Society.

Similarly to EBL, photolithography is a powerful technique for indirectly microstructuring MHPs, offering high throughput and spatial resolutions down to the sub-micrometer range, depending on the light source and process employed. However, as in the case of EBL, direct integration with MHP requires specifically tailored processing conditions, including careful control of soft-bake temperature and duration, as well as compatibility of the resists, developers, and removers with MHP materials.

Orthogonal photolithography-based MHPs microstructuring was reported for spin coated MAPbI₃ thin films by exploiting a negative photoresist (OSCoR1000) and highly fluorinated hydrofluoroether solvents.²⁶ Processing conditions include soft and post-exposure bake at 90 °C, exposure with 365 nm light (90 mJ cm⁻² dose) and development in 3-methoxyperfluoro(2-methylpentane) (HFE7300) for 5 minutes. The obtained pattern is then transferred to the MHP film by RIE with SF₆ plasma, transforming the unexposed MHP areas to PbF₂. Finally, resist stripping is performed in a mixture of methyl nonafluorobutyl ether and methyl nonafluoroisobutyl ether (HFE7100), resulting in a pattern with features down to 2 µm, as shown in Fig. 1C, with residual thickness of PbF₂ around 200 nm for a 350 nm thick MHP film. Both the developer and the remover were evaluated for their impact on the MHP film and did not show detectable damage after up to 10 hours of immersion.

Following this initial demonstration, several studies reported alternative microstructuring strategies based on photolithography employing various combinations of resists, developers and etching chemistries. Among these, the use of spin coated PMMA as a protective layer followed by standard photolitography processing was widely investigated.^17,18 Such an approach was recently reported for microstructuring spin coated 3D (CsPbBr₃, CsPbCl₃, MAPbI₃), quasi 2D (PEA₂ (MAPbBr₃)_(n−1)PbBr₄) and 0D (FAPbBr₃ nanocrystals) MHP films, using AZ MIR 701 as a positive photoresist on top of a PMMA protective layer, with results shown in Fig. 1D.¹⁷ After UV exposure, the patterns were developed and rinsed using water-based solutions, with the PMMA layer effectively encapsulating the underlying MHP. Reactive ion etching was then employed to simultaneously etch the exposed PMMA and the thin MHP layer, using HBr and BCl₃ for inorganic MHPs, and CF₄, Cl₂, and He for hybrid organic–inorganic MHPs. The remaining resist stack was subsequently removed by immersing the samples in hot toluene (80 °C for 1 h), yielding microstructures with feature sizes down to 1 µm. Post-etching optical and morphological analyses revealed material-dependent effects. For 3D perovskites, the film roughness increased, whereas a smoothing effect was observed for 2D and 0D films. The photoluminescence quantum yield (PLQY) remained unchanged for CsPbBr₃, but decreased for all other tested MHPs. In the case of MAPbI₃, compositional changes were identified at the structure edges, consistent with halide exchange in the plasma leading to the formation of MAPbI_(3−x)Cl_x. All other materials exhibited uniform emission across the patterned regions.

A similar approach was also demonstrated using a well-known negative photoresist (SU-8) on top of a PMMA layer, protecting the underlying MHP film from degradation upon contact with resist solvents (cyclopentanone and photoacids within the mixture) during spin coating.²⁷ In this case, after UV irradiation and development in ethyl lactate, the exposed pattern was transferred to the thin MHP film by first etching the PMMA protecting layer with an oxygen plasma, followed by MHP etching by argon ion milling. The remaining resist stack was finally removed by immersion in chloroform. An overall decrease on the PLQY, from 55% to 40%, was observed for this approach when applied to a (PEA)₂(MA)₂Pb₃Br₁₀ MHP, with the main drop happening upon UV irradiation, due to the limited photostability of the employed MHP. This microstructuring technique was further employed to fabricate dual-color micro-patterns, as shown in Fig. 1E, by combining two different MHP compositions. In this case, two distinct strategies were demonstrated, yielding different multicolor configurations. In the first approach, following the initial microstructuring step, the encapsulation of the patterned features with oversized PMMA and SU8 microstructures enabled the realization of multicolor patterns on the same substrate surface (Fig. 1E, left panel). Alternatively, PMMA and fully inverted SU8 layers deposited onto a previously patterned sample can serve as a secondary substrate plane for patterning the second-color microstructures (Fig. 1E, right panel).

Direct milling approaches

An alternative top–down approach for nanostructuring perovskites relies on surface milling by high-energy Ga⁺ ions bombardment in a focused-ion-beam (FIB) system. In this case, all issues associated with solvent interactions are completely eliminated, allowing the fabrication of nanometer-scale features; however, this comes at the cost of ion-induced surface degradation effects, including ion implantation, defects generation, and localized amorphization.^28–30 FIB milling was exploited for fabricating circular, linear and two-dimensional gratings directly into MHPs (Fig. 2A–C), including MAPbBr₃ crystals,^31,32 spin coated polycristalline MAPbI₃^33,34 and CsPbX₃ microcrystals.^29,35 Etching depth can be controlled by exposure dose, dwell time and writing loops. Interestingly, gas-assisted FIB milling was also investigated, where direct ablation from Ga⁺ ions is modified by chemical reactions with a gas precursor injected into the FIB chamber.^31,32 I₂ and XeF₂ alogen-based gases were tested, providing an additional degree of freedom for process control, with modified etching rates, surface damage and material re-deposition dynamics. Most importantly, this approach enabled to minimize direct effects on the MHP optical properties, partially recovering photoluminescence (PL) quenching upon exposure to Ga⁺ ions. More recently, cryogenic FIB milling was demonstrated to preserve the atomic structure of single crystal MAPbI₃ perovskite.³⁶ Although not yet utilized for direct nanostructuring, this approach has the potential to provide a cleaner access to nanostructured MHPs with preserved optical properties.

Fig. 2. (A) Linear and circular grating directly printed into perovskite MAPbBr₃ crystals by FIB XeF₂ gas-assisted etching. Adapted with permission from ref. 31 Copyright 2015, AIP Publishing (B) two dimensional nanoslit MHP metasurface obtained by FIB milling in polycristalline MAPbI₃. Groove width is around 100 nm.³³ Adapted with permission from ref. 33 Copyright 2017, Wiley-VCH. (C) Two dimensional photonic crystal lattice carved by FIB in a CsPbBr₃ microplatelet. Adapted with permission from ref. 35 Copyright 2024, Springer Nature. (D) Direct laser writing of microdisks in a MAPbI₃ thin film by using a donut-shaped fs laser beam. Adapted with permission from ref. 37 Copyright 2019, American Chemical Society. (E) Direct laser writing of a linear grating on the surface of a CsPbBr₃ nanowire by scanning a stripe-shaped femtosecond laser beam. Adapted with permission from ref. 38 Copyright 2021, American Chemical Society. (F) Micro-disk carving in CsPbBr₃ crystals by focused Gaussian laser beam writing. Spectra and microscope images inset show that structure cleaning after process in an ultrasonic bath is fundamental to obtain good edges and clean lasing action. Adapted with permission from ref. 39 Copyright 2023, IOP Publishing Ltd.

As an alternative direct microstructuring technique, ultrashort femtosecond laser ablation has been shown to provide high resolution and accuracy. Large scale microdisk laser lattice was defined on a spin coated 760 nm thick MAPbI₃ film on glass, exploiting an optical vortex laser beam (515 nm, 220 fs, Yb:KGW-based laser system) with donut-shape intensity profile, as shown in Fig. 2D.³⁷ The fabrication process delivered 1 cm² pattern in 15 minutes using a 200 KHz laser repetition rate and 100 mm s⁻¹ scanning speed, with microdisk spacing as small as 0.77 µm. The fabrication outcome can be precisely controlled by adjusting both the laser fluence, the vortex topological charge and the number of pulses applied, and microlasers with diameter from 2 to 9 µm were obtained. Optimal results in terms of surface and side-walls morphology are obtained using multiple pulses at fluences of a few tens of mJ cm⁻²—well below the single-pulse ablation threshold—yielding clean, smooth microstructures with minimal thermal damage to the MHP bulk. Single-pass laser ablation was used in combination with a tightly focused stripe-shaped laser beam (10 µm long, 400 nm wide, 515 nm wavelength with 180 fs pulses) to imprint 1D linear gratings on the surface of CsPbBr₃ nanowires, imprinting gratings with period down to 200 nm as shown in Fig. 2E.³⁸ Laser microstructuring with focused Gaussian beams was also demonstrated using a 520 nm Ti:sapphire femtosecond laser, employing a comparable low-power, multi-cycle strategy to produce microdisks with clean, well-defined edges in CsPbBr₃ single crystals, as sketched in the left panel of Fig. 2F.³⁹ Samples fabricated with this method required additional post-processing to remove residues generated during local ablation, including ultrasonic cleaning in toluene followed by annealing at 200 °C in an inert (N₂) atmosphere, finally resulting in clean lasing emission, shown in the right panel of Fig. 2F.

Scanning probe lithography (SPL) techniques offer an additional, though still underexplored, strategy for direct nanostructuring of MHP thin films. Mechanical SPL was recently demonstrated for nanostructuring 1D grating in 2D (PEA)₂PbI₄ thin films.⁴⁰ The method relies on directly scanning an atomic force microscope tip along a predefined trajectory while precisely controlling its distance from the sample surface, thus carving well-defined trenches into the MHP film. The depth of the trench was adjusted by tuning the cantilever force in the µN range, whereas the edge geometry was improved by employing slow scan speeds of approximately 1 µm s⁻¹. The resulting trench profile was strongly influenced by the AFM tip geometry: using conical tips and operating at depths of 15–50 nm yielded widths of 80–130 nm at half-height. Although promising for the direct fabrication of larger structures such as waveguides and couplers, the technique displaces material across the MHP surface, leaving partial residues along the edges and at the trench termini. As with direct laser writing, additional post-processing may therefore be required to obtain smoother, better-defined structures.

Micro- and nanoimprint lithography

Micro- and nanoimprint lithography (MIL/NIL), as top–down patterning approaches, have recently been extended to MHP materials for the geometric definition of micro- and nanostructures.²⁰

In general, imprint lithography offers several advantages in micro- and nanofabrication by employing a simple and efficient pattern-transfer process. In this approach, a mold containing micro- or nanoscale features is pressed into a deformable or liquid material deposited on a substrate, after which the pattern is solidified by thermal or ultraviolet curing prior to mold removal.^41,42 Those techniques enable the fabrication of extremely high-resolution structures, often surpassing the limits of conventional photolithography. Moreover, imprint lithography is cost-effective as it eliminates the need for complex optical systems, supports large-area and high-throughput processing, and is compatible with a wide range of materials, including MHPs, making it well suited for large-scale manufacturing.⁴³

Thermal nanoimprint technique applied to MHPs was first reported in 2016 to realize 2D photonic-crystal lasers.²⁰ Preformed silicon molds (flat, nanopillar, or nanograting), were treated with antiadhesion 1H,1H,2H,2H-perfluorodecyltrichlorosilane monolayer then imprinted onto pre-deposited MAPbI₃ MHP films under identical conditions (100 °C, 7 MPa, 20 min). The influence of nanoimprint pattern geometry on photodector performance was then examined, reporting promising performance. The technique was further applied to fabricate large-area Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃ MHP metasurfaces.⁴⁴ The MHP nanostripes and nanoholes precisely replicate the inverse patterns of silicon molds, with diameters and periodicities of roughly 300 nm and 600 nm, respectively (Fig. 3 A(i) and (ii)). The obtained metasurfaces operate in visible and near-infrared ranges, and exhibit significantly enhanced PL (Fig. 3A(iii)). Importantly, the Cs_0.05 (MA_0.17FA_0.83) _0.95Pb(I_0.83Br_0.17)₃ alloy exposed for 3000 h of lying in ambient conditions preserves PL properties.

Fig. 3. (A) Schematic illustration of nanoimprint lithography employing a silicon mold with (i) nanostripe and (ii) nanohole geometries, together with SEM images of the corresponding patterned Cs_0.05(MA_0.17FA_0.83)_0.95Pb(I_0.83Br_0.17)₃ MHP metasurfaces. (iii) Conceptual diagram depicting photoluminescence enhancement in the MHP metasurfaces mediated by Mie resonances. Adapted with permission from ref. 44 Copyright 2017, American Chemical Society. (B) (i) Schematic representation of the polymer-assisted nanoimprinting with polymer additive fabrication process. A CsPbX₃ (X = Br or I) precursor solution blended with PEO is first deposited by spin coating to form an intermediate adduct layer. The film is then pressed with a prepatterned PDMS mold under conformal contact at approximately 100 kPa (Step I), followed by thermal treatment at 80 °C for 15 min to induce crystallization (Step II). After demolding (Step III), well-defined CsPbX₃ (X = Br, I) nanopatterned structures are produced. (ii) Optical microscopy image of a CsPbBr₃ nanopatterned film, highlighting the surface fringe features; an inset photograph provides a macroscopic view of the fringe pattern. (iii) Field emission FE-SEM images of representative nanopatterned structures which can be obtained via polymer-assisted nanoimprinting, including square-hole arrays with feature widths of 500 nm or 300 nm arranged in P4mm symmetry, circular holes with a diameter of 200 nm, and rectangular apertures with dimensions of 200 nm × 2 µm. Adapted with permission from ref. 45 Copyright 2020, American Chemical Society.

In general, in NIL processes it is of paramount importance to control the pattern quality during thermal/optical treatment along with the control over the filling process: material properties and pattern size and shape play a crucial role in this sense. In the specific case of MHPs, and more specifically in fully inorganic MHPs, achieving reliable nanoimprinting requires a precursor film with adequate softness and deformability. It is also crucial to suppress rapid crystallization, so that the film remains in an amorphous adduct state long enough for a pattern transfer with high fidelity and resolution. To solve this problem, polymer-assisted nanoimprinting using poly(ethylene oxide) (PEO) additive was developed to produce submicrometer MHP nanopatterns on different substrates (Fig. 3B(i)).⁴⁵ The inclusion of a tiny amount of PEO allowed the fabrication of soft and moldable adduct films by delaying MHP crystallization in the adduct state, providing sufficient processing time to imprint the film at room temperature under gentle pressure using a prepatterned polydimethylsiloxane (PDMS) stamp. Using this strategy, well-defined inorganic MHP nanopatterns of CsPbBr₃ and CsPbI₃ of 200 nm width and extended over a large area, have been demonstrated (Fig. 3B (ii)). The imprinted CsPbI₃ line arrays were subsequently employed in two-terminal parallel photodetection devices. Photoactive CsPbBr₃ and black-phase CsPbI₃ were patterned at the hundred-nanometer scale into diverse periodic geometries, including line arrays, square motifs, hexagonally arranged apertures, and rectangular features (Fig. 3B(iii)). To protect the as-fabricated CsPbI₃ nanopatterns, a polymer backfilling approach was developed, involving overcoating a thin poly(vinylidenefluoride-co-trifluoroethylene) (PVDF-TrFE) film approximately on top of the CsPbI₃ nanopatterns fabricated followed by thermal processing. This method preserved the photoabsorption stability of CsPbI₃ for up to one year under ambient conditions and for 10 hours at 85 °C. A further advancement was recently reported, in which the inclusion of self-assembled block copolymer, poly(styrene)-block-poly(2-vinylpyridine) (PS-b-P2VP), into MHP precursors combined with NIL allowed the formation of hierarchically controlled perovskite nanostructures. These structures ranged from molecular-scale crystals to self-assembled domains on the order of tens of nanometers, as well as patterned features with dimensions of several hundred nanometers.⁴⁶ The study further shows that nanoimprinting forms an ultrathin PS surface layer on the nanostructures, arising from the preferential affinity between the PS block and nonpolar PDMS mold. This enhances the environmental robustness of the resulting hierarchically controlled MHP nanostructures that maintain a PLQY of ≈28% over 100 days upon exposure to ambient conditions (25 °C, 30–50% relative humidity).

Thermal NIL was exploited to realize 2D photonic-crystal lasers on MAPbI₃ polycrystalline films, at temperatures as low as 100 °C. The nanoimprint stamps consisted of a 500 µm-thick silicon backplane and an approximately 20 µm-thick patterned Ormostamp (Micro Resist Technology GmbH) top layer, comprising pillars arranged in a triangular lattice with a 450 nm pitch and a pillar diameter of 230 nm.⁴⁷ The controlled temperature and pressure during the imprint process allowed a faithful transfer of the stamp pattern to the MHP layer over a large area. The authors propose that imprinting at 100 °C promotes recrystallization of the polycrystalline MHP into the tetragonal phase without decomposition, simultaneously smoothing the surface and reducing propagation losses for optical modes within the layer. In addition, thermal imprint-induced recrystallization removes surface defects that would otherwise promote carrier recombination and luminescence quenching. As a result, 2D photonic-crystal lasers exhibit a low lasing threshold of 3.8 µJ cm⁻². Long-term stability tests of the imprinted laser structures yielded encouraging results, with lasing remaining stable for more than 8 hours under operation at 15% above threshold and a 1 kHz repetition rate.

Subsequently, a modified micro-imprint technique was developed to produce high-quality 1D perovskite triangular nanowire (TNW) array, exploiting a space-confinement strategy (Fig. 4A(i)–(iii)).⁴⁸ MAPbI₃ precursor solution was deposited, followed by vertical placement of a soft PDMS mold. Upon applying appropriate pressure, capillary forces guided the precursors' solution to uniformly occupy the confined region between the template and the substrate. The system was then heated to 100 °C, promoting preferential nucleation at high-energy sites along the nanochannel edges closest to the heat source. As the solvent gradually evaporated, crystallization proceeded under nanochannel confinement, yielding highly crystalline TNW arrays. The hydrophobic surface, conformal contact, and mechanical flexibility of PDMS preserved the structural integrity and high quality of the transferred TNWs, which was then assessed with integration into actual devices. Photodetectors fabricated using the resulting MAPbI₃ TNWs demonstrated indeed outstanding photoresponse performance. A similar method was developed to produce enlarged, preferentially oriented grains that reduce nonradiative recombination losses in solar cells.⁴⁹ Using a polyurethane stamp with hexagonal nanodot arrays, applied under controlled pressure on partially dried MHP intermediate films, induced pressure-driven crystallization, resulting in preferred-orientation and enlarged grains with low-angle grain boundaries.

Fig. 4. (A) High-quality 1D MHP triangular nanowire array synthesis through space-confined effect. SEM (i) plane view, (ii) side view images of the single crystal MAPbI₃ triangular nanowire arrays. (iii) Representative AFM micrographs in 2D and 3D MAPbI₃ arrays. The inset shows the corresponding optical photographs. Adapted with permission from ref. 48 Copyright 2023, Wiley-VCH. (B) MHP single-crystal arrays fabricated by dry imprint transfer. (i) Schematic workflow illustrating the dry imprint transfer process. Large-area monocrystalline CsPbBr₃ and CsPbCl₃ films are firstly epitaxially grown on muscovite mica via CVD, then a thermal release tape (TRT) is used to transfer MHP films to flat PDMS that work as buffer substrate for stamping and mechanical imprinting. Stress is applied with customised SiO₂ stamper continuously until the MHP film reaches its fracture limit and detaches along the stamper edges. The punched units are eventually picked up with TRT at room temperature and transferred to the target substrate. (ii) Optical microscopy image of the transferred arrays on the target substrate. (iii) Optical microscopy image of array with square, triangular, and circular shapes subsequently deposited in the same position (iv) and CsPbBr₃–CsPbCl₃ vertical heterostructure array. Adapted with permission from ref. 50 Copyright 2025, American Chemical Society.

An evolution of thermal imprint lithography has recently been reported.⁵⁰ A dry mechanical patterning method was developed using rigid SiO₂ stampers with predefined geometries to mechanically punch monocrystalline MHPs (CsPbBr₃ and CsPbCl₃), which were pre-synthetized via CVD (Fig. 4B(i)). This allowed the fabrication of MHP single-crystal arrays while preserving their intrinsic material properties. This imprint process is then followed by dry elastic stamp transfer to the desired substrates. Thanks to this approach, the patterned MHP single crystals exhibited high uniformity and minimal processing-induced damage, as it avoids solvents, photoresists, and etching processes that typically degrade MHPs (Fig. 4B(ii)).

The flexibility of the dry imprinting transfer approach is highlighted by its capability to construct complex architectures relevant to optoelectronic integration. As illustrated in Fig. 4B(iii), the process allows sequential imprinting transfers at identical locations, enabling the formation of stacked arrays composed of square, circular, and triangular CsPbBr₃ elements. By leveraging sequential imprint transfer, together with the capability to pattern arrays of different MHP compositions, the authors fabricated an array of CsPbBr₃–CsPbCl₃ heterostructures (Fig. 4B(iv)). The application of this technique was further demonstrated fabricating photodetector arrays.

Overall thermal MIL/NIL has been applied in several fashions to different MHP compositions (from hybrid to fully inorganic) and to polycrystalline and single crystalline materials. These patterning approaches have been demonstrated to be powerful tools for tailoring MHP materials into designed geometries without inducing structural damage while in some cases enhancing their photophysical and optoelectronic properties.

In summary, top–down fabrication of MHPs encompasses three main strategies: mask-mediated indirect lithography, direct milling, and micro-/nanoimprint lithography, each offering distinct trade-offs between precision, scalability, and material compatibility. Indirect approaches, such as electron-beam lithography and photolithography, are well established and provide high resolution down to the nanometer scale for the former and the sub-micrometer range for the latter depending on the light source and process conditions, along with considerable geometric design flexibility. However, the required chemical or physical etching steps may lead to significant degradation of MHP materials. Direct milling techniques, including FIB processing, laser ablation, and SPL, enable resist-free structuring and relatively simple workflows, each offering distinct trade-offs in resolution, possible material damaging, and scalability. Ultrashort femtosecond laser ablation enables direct microstructuring with high precision, although additional post-processing is often required to remove residues generated during local ablation. Similarly, SPL can mechanically define nanostructures in MHPs, but may leave partial residues along the produced structures and, like direct laser writing, can require further post-processing. In contrast, FIB provides nanoscale accuracy but may induce ion-related damage, including ion implantation, defect formation, and localized amorphisation, while also suffering from limited scalability. Micro- and nanoimprint lithography complements these approaches by offering a versatile and cost-effective route for large-area, high-throughput fabrication, capable of producing high-resolution nanometer-scale structures while often preserving or even enhancing MHP properties. Overall, the optimal top–down strategy depends on balancing structural fidelity, processing complexity, and the intrinsic sensitivity of MHP materials, with each technique being best suited to specific device architecture and performance requirements. Complementary to the aforementioned top–down fabrication strategies, bottom–up direct growth approaches such as microfluidics and dewetting-assisted growth are gaining increasing attention. These methods follow a fundamentally different route by forming the desired micro- and nanostructures in situ during crystallization, while still enabling a high degree of structural control and compatibility with device integration. As an example, microfluidics-assisted growth exploits fluid dynamics, wettability contrasts, and confined geometries to precisely control nucleation, crystal growth, and spatial arrangement, thereby helping to preserve the intrinsic optoelectronic properties of MHPs. Geometrically confined crystallization, for instance, relies on polymeric molds or silicon templates to shape precursor solutions into predefined microstructures, generally maintaining crystal quality more effectively. These techniques are discussed in greater detail and critically examined in the following paragraph.

Direct-growth approaches

The overall performance of optoelectronic devices strongly depends on the structural arrangement of active materials, as exemplified by hybrid MHPs, where a well-established correlation exists between microstructure and photophysical properties in optoelectronic applications.^51,52 In recent years, fabrication methods based on the self-assembly of building blocks driven by liquid-phase processes have proven to be strategic for creating functional micro- and nanostructures or arrays. These methods can be broadly classified into two main categories, depending on the dynamic and fluid transport processes involved: dewetting-assisted growth and microfluidics-assisted growth. In addition to these approaches, structured MHPs can be grown via conventional lithography and lift-off, in which a resist mask defines the areas where the MHPs develop. Direct-growth approaches enable precise control over the size, shape, and composition of structured MHP as well as their position across the substrate, making them highly compatible with device integration. Furthermore,compared with post-growth micro- and nanostructuring methods, they can reduce the risk of processing-induced damage, thereby helping to preserve the intrinsic optoelectronic properties of MHPs.

Dewetting-assisted growth

The exploitation of surfaces with wettability contrast is an effective strategy, widely observed in nature, to guide and confine fluids. Owing to the solution processability of MHPs and the strong dependence of the crystallization process on equilibrium conditions in MHP precursors, several approaches for the fabrication of MHP micro- and nanostructures rely on the precursors assembly on surfaces with wettability differences. The behavior of liquids on solid surfaces results from the complex interplay between multiple interaction interfaces, including the liquid–solid, liquid–vapor, and solid–vapor boundaries.^53–56

The region where the solid, liquid, and gas phases meet is defined as the triple contact line (TCL). The associated interfacial energies determine the meniscus shape, solvent evaporation, supersaturation, and finally crystallization. Variations in surface tension or wettability introduce controlled local perturbations and instabilities that can lead to spreading, retraction, or pinning of the TCL, directly affecting solute transport, supersaturation, and crystallization.

Substrates characterized by a wettability contrast between different regions can effectively confine the precursor solution. The solution spontaneously migrates toward areas with higher wettability, filling the hydrophilic regions where nucleation occurs. In correspondence with these hydrophilic areas, the TCL becomes pinned, ensuring uniform solvent evaporation and localized crystal growth.

Exploiting this wettability-mediated confinement, Lee et al.²³ demonstrated the fabrication of MHP nanoplate arrays using a substrate patterned with wettable square regions. These nanostructures were obtained through selective functionalization with hexamethyldisilazane and 1H,1H,2H,2H-perfluorodecyltriethoxysilane, yielding relatively hydrophobic and superhydrophobic regions, respectively. The precursor solution was spread across the substrate using a shearing blade. Because of the wettability contrast, the contact angle of the precursor solution and the length of the solution meniscus varied across the surface, with lower (higher) contact angles and longer (shorter) meniscus lengths in the hydrophilic (hydrophobic) regions. This behavior ensured effective wetting and crystallization within the hexamethyldisilazane-patterned areas, resulting in platelets with lateral dimensions of a few microns (Fig. 5A).

Fig. 5. (A) Fabrication of 2D MHP nanoplate array using a wettability pattern and a shearing process (top), with corresponding PL image of the nanoplate array (bottom) and SEM image of a single crystal. Adapted with permission from ref. 23 Copyright 2023, American Chemical Society. (B) Evaporation dynamics and hydrodynamic flows in a DMSO/CHP precursor droplet and resulting crystal micrograph. Light violet, green, and light blue arrows represent solvent evaporation, capillary flow, and Marangoni flow, respectively. Adapted with permission from ref. 57 Copyright 2025 Wiley-VCH. (C) SEM images of crystals grown in square and triangular wells (top) and the corresponding meniscus shapes (bottom) highlighting the crystal positioning control. Adapted with permission from ref. 58 Copyright 2023 Springer Nature. (D) Capillary bridge growth of (101)-oriented 2D MHP nanowire crystallization in a capillary bridge and PL image of the wires during growth. Adapted with permission from ref. 59 Copyright 2020, Wiley-VCH. (E) Schematic of the capillary corner illustrating directional mass transport, nucleation, and nanowire array formation, along with corresponding PL image. Adapted with permission from ref. 22 Copyright 2025, American Chemical Society.

In another recent work, Zhang et al.⁵⁷ developed an approach that combines a static wettability contrast with a surface tension gradient to successfully synthesize MHP subwavelength nanostructures with high crystallinity and well-defined shapes. The static wettability contrast was achieved by patterning the growth substrate with heptadecafluorodecyltrimethoxysilane through an initial functionalization step, followed by exposure to oxygen plasma in the presence of a metal mask. This process enabled the formation of hydrophilic squares surrounded by highly hydrophobic regions. As in the previous cases, this strategy ensured confinement of the precursor solution at specific locations on the substrate, allowing precise microplatelets formation. To further improve crystal quality, the authors introduced a surface tension gradient through the appropriate selection of a binary solvent mixture with different boiling points, surface tensions, MHP precursor solubility, and viscosity. Most importantly, the difference in boiling point and surface tension between the two solvents enabled the formation of a surface tension gradient induced by their differential evaporation within the binary mixture. This effect can be achieved, for example, by adding 15–25% of 1-cyclohexyl-2-pyrrolidone (CHP) to a MHP precursor solution in dimethylsulfoxide (DMSO). Because of the lower boiling point of DMSO, its faster evaporation causes CHP to accumulate at the TCL, reducing the surface tension and contact angle. This established a surface tension gradient from the center of the droplet toward the TCL, inducing an enhanced Marangoni flow toward the center (Fig. 5B). This flow promoted the growth of nuclei formed at the droplet center while suppressing additional nucleation events. In this way, the coffee-ring effect was suppressed, enabling the formation of single crystals with high crystalline quality. Furthermore, this approach can be combined with a space-confined crystallization method. A rigid stencil patterned with the desired structures is placed on top of the precursor solution once nucleation has started. This enables the formation of arrays of single crystal microplatelets structured with subwavelength nanostructures. When these platelets are incorporated into photodetectors, they exhibit promising performance and high switching stability, maintaining similar photocurrent values even after up to 15 000 on–off cycles. Another approach that exploits surface wettability to synthesize nanometric crystals at specific substrate sites uses templates with wells of defined shapes, fabricated through EBL in a hydrogen silsesquioxane (HSQ) resist film.⁵⁸ The inner sides of the wells and the HSQ surface exhibited different wettabilities, with the surface being highly hydrophobic compared to the wells. This contrast was achieved through a two-step functionalization with fluorinated silane molecules of different chain lengths, yielding higher hydrophobicity on the surface and lower hydrophobicity within the wells. The well geometry played a crucial role in determining the droplet meniscus shape, influencing solvent evaporation and, consequently, the final position of the crystals, which form where the meniscus thickness is higher, typically near the well corners. For wells with symmetric geometries, such as squares or equilateral triangles, the symmetric meniscus allowed crystals to form randomly at any corner. In contrast, asymmetric wells, like isosceles triangles, induced the formation of an asymmetric meniscus, localizing the nanocrystals at the corner with the smallest angle (Fig. 5C). Using this strategy, it was possible to obtain crystals with dimensions on the order of tens of nanometers and with extremely high positional accuracy, a key requirement for fabricating arrays for applications in miniaturized optoelectronic devices.

As an alternative approach, capillary bridge-mediated assembly allows the growth of ordered MHP micro- and nanostructures by exploiting transient capillary bridges formed between a liquid film and a patterned template. In this method, the template structures exhibited asymmetric wettability, with hydrophilic tops and hydrophobic sidewalls, which ensured that the polar MHP precursor solution was confined to the tops of the structures, promoting conformal crystallization. As solvent evaporation proceeded, curvature and pressure gradients within the meniscus generated localized capillary forces that drove material accumulation and induced highly ordered nucleation at predefined sites (Fig. 5D). Dynamic control of humidity, temperature, and solvent volatility allowed precise tuning of bridge lifetime and crystal morphology, enabling the fabrication of well-defined architectures. Using this strategy, several works achieved MHP microstructures with feature sizes typically ranging from 1 to 10 µm and periodicities up to several tens of micrometers, including highly uniform arrays of microplates and microwires. MHP nanowire arrays, in particular n = 3 quasi-2D perovskites, consisting of three layers of inorganic octahedra separated by large organic cations, prepared with this approach have demonstrated excellent stability under ambient conditions (25–30 °C and 40% relative humidity). No variation in the crystal structure or optical properties is observed, as confirmed by unchanged diffraction patterns and PL spectra after 30 days of ambient exposure.^59–63

A variation of this approach is the capillary corner bridge, which confines the MHP precursors solution at the edges of the template structures, enabling the crystallization of subwavelength MHP nanowire single crystals.²² This method relied on a precise modification of the wettability of both the template and the substrate. Unlike the classic capillary bridge approach, the tops of the template structures were functionalized to be superhydrophobic, while the sidewalls exhibited lower hydrophobicity. This was achieved through a two-step functionalization using fluorinated alkyl chain silane molecules of different chain lengths. Simultaneously, the hydrophobicity of the target substrate is tuned with octadecyltrichlorosilane, allowing wettability to be adjusted depending on the treatment duration.

During solvent evaporation, the modified micropillar template induced the formation of square capillaries of precursor solution at the micropillar corners, driving solute mass transport toward the tip of the capillary. As a result, supersaturation is rapidly reached, favoring nucleation at the capillary corner menisci (Fig. 5E). This approach also allowed for precise tuning of the nanowire geometry. For example, nanowire size can be controlled by adjusting the contact angle of the target substrate and template sidewalls or by varying the precursor solution concentration, while the spacing between wires was determined by the width of the micropillar template. This method demonstrated versatility and significant potential for the development of advanced optoelectronic and photonic devices with high stability. Indeed, stability tests performed on subwavelength MHP nanowires and corresponding photodetectors after 60 min of irradiation show no changes in crystal quality, as confirmed by unchanged X-ray diffraction (XRD) patterns measured post-irradiation. Moreover, the photodetectors maintain stable photocurrent under the same irradiation conditions, indicating preserved performance.

Overall, these results demonstrate that exploiting wettability contrasts provides precise control over the assembly of molecular MHP precursors, enabling the formation of high-quality micro- and nanostructures. Indeed, by modulating spatially confined and supersaturated regions, it is possible to control interfacial instabilities, and effectively guide the organization of building blocks.

Microfluidics-assisted growth

A versatile and effective approach to obtaining shaped MHP microcrystals involves the use of microstructured templates, which can be rigid (for example, porous anodized aluminum oxide) or flexible (for example PDMS). These templates serve to confine the MHP precursor solution and direct crystal growth, ultimately determining the morphology.

Among all, PDMS molds are a popular choice due to their flexibility and transparency, and the Micromolding in Capillaries (MIMIC) technique is one of the most widely used strategies. Initially introduced by Whitesides et al.,⁶⁴ this method involves pressing an elastomeric PDMS stamp with a patterned surface onto a substrate, thereby creating a network of microchannels. A liquid precursor solution or suspension fills the channels by capillary action, where it subsequently cures or dries to form MHP microstructures complementary to that of the mold. Finally, the stamp is removed, leaving the desired microstructured material on the substrate.

This technique was originally employed to fabricate microstructures of organic and inorganic materials, and it is now widely used to produce high-quality MHP crystals with controlled thickness and shape.

One of the earliest examples of the application of this technique, using rectangular PDMS groove templates, demonstrated the synthesis of arrays of MAPbX₃ MHP nanowires²¹(Fig. 6A). The process enabled precise control over the widths (ranging from 460 to 2500 nm) and lengths (from 10 to 50 µm) of individual nanowires, as well as over the wire-to-wire spacing. Furthermore, the thickness of the nanowires (between 80 and 1000 nm) was tuned by adjusting the concentration of the precursors solution. It was observed that, following the slow evaporation of the solvent N,N-dimethylformamide (DMF), occurring over 2–3 hours, small crystalline domains first appeared at the ends of the templates. Once these primary nuclei were formed, they rapidly grew (<20 s) along the length direction of the PDMS mold.

Fig. 6. (A) Crystal growth process in PDMS template and morphology of MAPbI₃ perovskite nanowires arrays. Adapted with permission from ref. 21 Copyright 2017, American Chemical Society. (B) Sketch of template-confined, microfluidics-assisted growth approach and SEM image of a Dion-Jacobson MHP microwire. Adapted with permission from ref. 67 Copyright 2025, Wiley-VCH. (C) Schematic of the PDMS template-confined antisolvent crystallization method and optical image of CsPbBr₃ microwires arrays. Adapted with permission from ref. 68 Copyright 2019, Wiley-VCH. (D) Synthesis of MHP nanowire arrays in Anodized Aluminum Oxide (AAO). Adapted with permission from ref. 69 Copyright 2016, American Chemical Society. (E) Sketch of the blade-coating method for MHP microwire array preparation. Adapted with permission from ref. 70 Copyright 2025, Wiley-VCH.

Using the same method, arrays of CsPbX₃ (X = Cl, Br) microdisk lasers with a rectangular geometry and side lengths of approximately 2.5 µm were also fabricated by employing PDMS templates featuring cylindrical holes.⁶⁵

The versatility of this technique also enabled the fabrication of single crystals of quasi-2D (BA)₂(MA)_n−1Pb_nBr_3n + 1 (with n = 4) MHP.⁶⁶ Microwires of different widths (8 µm, 15 µm, 30 µm, and 100 µm) were obtained by properly designing micro-sized periodic strip-groove PDMS templates. The authors systematically investigated how the process parameters influence crystal quality, providing useful insight for the control of the crystallization process. They found that a MHP precursor solution in DMSO with a concentration of 30 wt% was optimal for obtaining high-quality crystals. Lower concentrations led to discontinuous microwires due to insufficient solute content for crystal formation, whereas higher concentrations caused excessive lateral crystal growth, resulting in the formation of residual layers between adjacent microwires. In addition, temperature was identified as a crucial parameter for crystal growth. Although higher temperatures accelerated the process, excessively high values caused rapid evaporation of DMSO, which could deform the PDMS template and disrupt the controlled crystal growth. Consequently, the morphology was no longer governed by the template, leading to disordered structures instead of continuous and uniform microwires. To ensure high crystal quality, the optimal temperature range was found to be between 50 and 80 °C. Moreover, the devices fabricated using these microwires exhibited remarkable stability, showing no significant change in performance after being exposed to an environment with 60% relative humidity for several days.

Following these general guidelines and using the approach illustrated in (Fig. 6B), a wide variety of structured MHPs has been synthesized, including 3D,^71–73 2D,⁷⁴ and Dion–Jacobson-type MHPs.⁶⁷ In this process, a patterned PDMS template is first placed onto the substrate, the precursor solution is then injected into the capillaries, where solvent evaporation induces MHP precipitation; finally, after removal of the template, single-crystal microwires are obtained on the substrate. The resulting structures can be tailored by carefully tuning the precursor concentration, solvent, and reaction temperature according to the desired microstructure dimensions.

Moreover, more complex architectures were also fabricated, such as MAPbBr₃–MAPbI₃ heterostructures syntesized via a sequential template-assisted approach.⁷⁵ After the growth of MAPbBr₃ microwire arrays on a Si/SiO₂ substrate, a PDMS template with periodic rectangular cavities filled with a MAPbI₃ precursor solution in DMF was aligned with the preformed MAPbBr₃ arrays along the same orientation. To avoid MAPbBr₃ damage, an optimized MAPbI₃ precursor composition was employed, and the reaction was conducted at 80 °C, a temperature at which MAPbBr₃ shows low solubility in DMF. Furthermore, given the importance of finely controlling crystallization kinetics, the microfluidic synthesis approach was properly combined with additional strategies, namely antisolvent-assisted crystallization and accelerated solvent evaporation.

In the case of the antisolvent-assisted crystallization, acetonitrile (CH₃CN) was used both to induce saturation of the DMSO precursor solution and to saturate the growth environment, thereby promoting rapid and efficient crystallization (Fig. 6C). Monocrystalline CsPbBr₃ microwire arrays with tunable widths (from 2 to 7 µm) and lengths of up to 100 µm were obtained⁶⁸ by placing the reaction system—consisting of the DMSO precursor solution saturated with CH₃CN confined between the PDMS template and the substrate—on a Teflon stage in a sealed CH₃CN atmosphere at 50 °C for 12 h, then refilling it once with fresh precursor solution without removing the template and repeating the process. The crucial role of the antisolvent was demonstrated by the observation that when the precursor solution was not presaturated with CH₃CN, only unidentified impurities were formed instead of high-quality single crystals.

On the other hand, in another report, accelerated evaporation of the DMF solvent from the MAPbBr₃ perovskite precursor solution was induced by heating the reaction system to 60 °C.⁷⁶ To prevent the PDMS template from being lifted off the substrate by the solvent vapor during evaporation, a glass spacer was placed at one end of the template, creating an opening that allowed the vapor to escape from the microchannels. Using this method, both linear and curved MAPbBr₃ single-crystal microwire arrays were successfully obtained by varying the template patterns. The dimensions of the microwires could be tuned by adjusting the precursor concentration: heights ranging from 95 to 750 nm and widths from 510 nm to 3.2 µm were achieved by changing the MHP precursor concentration from 5 to 23.5 wt%. Photodetectors based on this perovskite array exhibited impressive long-term stability, retaining 96% of their initial photocurrent after more than one year of exposure to air without additional encapsulation.

An advanced implementation of the accelerated solvent evaporation strategy allowed the fabrication of large-area, single-crystal MAPbI₃ MHP thin films via a roll-printing method with geometrically confined lateral crystal growth.⁷⁷ The resulting films exhibited continuous and highly uniform single-crystalline domains, hundreds of micrometers in width and 80–120 nm in thickness, seamlessly covering 2-inch wafers. In this process, a perovskite ink was transferred through a channel-patterned PDMS rolling mold onto a heated substrate (180 °C), where immediate solvent (DMF) evaporation induced rapid nucleation and unidirectional crystal growth under geometric confinement, resulting in large single-crystal domains.

In addition to flexible PDMS molds, rigid templates were also used, which generally were not removed at the end of the synthesis. As a result, the final product was a MHP embedded within a microstructure made of another material.

One of the first examples reported the use of Anodized Aluminum Oxide (AAO) to obtain uniform MHP nanowires (MAPbI₃, MAPbBr₃, and Cs₂SnI₆) with controlled diameters ranging from 50 to 200 nm (Fig. 6D).⁶⁹ The process involved adding a MHP precursor into AAO templates with oriented cylindrical nanopores, followed by MHP dissolution and recrystallization. The resulting nanowires were oriented perpendicular to the substrate (transparent conductive electrode made of TiO₂/FTO glass).

In addition, Si/SiO₂ microchannel arrays have also been used to obtain high-quality CsPbBr₃⁷⁸ and MAPbI₃⁷⁹ MHP microwires confined within the microchannel platform. To finely control the crystallization kinetics, a micro/nanopumping strategy was implemented. This setup consisted of a precursor solution in a Petri dish that act as the solution source, while a filter paper placed on a glass slide funcioned as a micropump, transferring the MHP solution from the source into the parallel microchannels of the substrate.⁷⁸

Moreover, a blade-coating process on periodic SU8 photoresist microchannels was implemented to control solution flow dynamics and thereby achieve uniform growth of MAPbI₃ microwire arrays.⁷⁰ In this process, when the precursor solution was loaded onto the microchannels and filled them via capillary action, a blade was used to drag the perovskite solution across the microchannel-covered substrate at a temperature of approximately 35 °C. As the solvent evaporated, continuous crystal growth occurred at a coating speed of 30 µm s⁻¹, producing MHP single-crystal arrays within the microchannels (Fig. 6E).

Physical mask-assisted growth

Lithography and lift-off approaches can be effectively exploited to define physical masks on top of target substrates, where MHPs grow in the regions defined by the mask, followed by mask removal through lift-off.

Using this approach, the direct growth of 2D MHPs nanoantenna lattices was demonstrated.⁸⁰ Here, a double layer PMMA resist was structured by EBL followed by PbI₂ evaporation. This thin layer is converted to MHP after exposure to MAI vapor in a tube furnace. Excess MHP and PMMA are finally removed by lift-off in toluene, leaving MAPbI₃ lattices of cylindrical nanoantennas with 300 nm diameter and 100 nm thickness on the substrate, as shown in Fig. 7A. A similar approach exploited EBL of a thin PMMA layer deposited on top of atomically thin PbI₂ nanosheets, as depicted in the left panel of Fig. 7B. In this case the standard PMMA developer (MIBK : IPA 1 : 3) was found not to damage the PbI₂ precursor. After development, the exposed areas were converted to MHP in a CVD process by exposure to MAI vapor. In such a way, nanostructured MHP wires with linewidth down to 150 nm were obtained, but maintaining the presence of PbI₂ in the unexposed areas.⁸¹ This approach enabled the fabrication of MHP with different halogen composition, and hence different photoemission spectrum, on the same optoelectronic device combining a multi-step growth process, as shown in the right panel of Fig. 7B.

Fig. 7. (A) MHP nanodisk lattices obtained by EBL on PMMA, followed by PbI₂ precursor evaporation and MHP growth in CH₃NH₃I vapors. Adapted with permission from ref. 80. Copyright 2023, The American Association for the Advancement of Science (B) Multi-color MHP patterning by EBL processing on PbI₂ nanosheets with a single layer PMMA. Scale bar is 10 µm and corresponding photoluminescence spectra are shown. Adapted with permission from ref. 81 Copyright 2022, Wiley-VCH. Multi-color patterning by photolithography and lift-off exploiting (C) a double layer PMMA and deep-UV lithography followed by liftoff and (D) parylene-C as a sacrificial layer followed by quasi-2D MHP precursors spin coating and peeling-off. (C) Adapted with permission from ref. 82 Copyright 2018, Wiley-VCH. (D) Adapted with permission from ref. 83 Copyright 2020, American Chemical Society.

Perovskite microstructuring based on photolithography and lift-off was also reported. One such approach was based on a double layer PMMA (50–950 K) resist and deep-UV (λ = 220 nm) photolithography, followed by thermal co-evaporation of MAI and PbI₂ to form structured MAPI₃ perovskite after lift-off in chlorobenzene, as shown in the top panel of Fig. 7C.⁸² With such an approach large area microstructuring was achieved, together with minimal affection of the PL spectrum between structured and unstructured areas (Fig. 7C, bottom panel). Thin film transistors were fabricated with such an approach, delivering operating devices with stable performance upon electrical, illumination, and mechanical stress, under ambient conditions. A second approach was also reported, shown in Fig. 7D, based on dry lift-off using photolithography of a negative photoresist (NR9-3000) structured on parylene-C film deposited by room temperature CVD.⁸³ After photolithography, RIE was employed to transfer the microstructures into the 2.5 µm-thick parylene-C layer, followed by spin-coating of the MHP precursors and annealing at 100 °C to promote crystallization. This method leverages the inherently low adhesion of parylene to the underlying substrates (glass or Indium Tin Oxide (ITO)-coated glass), allowing the film to be peeled off directly with tweezers and enabling the fabrication of structures as small as 4 µm. This strategy was also applied to generate multi-color patterns using quasi-2D MHP with different halide compositions on the same sample, by cascaded microstructuring processes.

Summarising, bottom–up direct-growth approaches, including dewetting-assisted, microfluidic assisted, and physical mask-assisted growth, provide a versatile and increasingly powerful route for the fabrication of shaped MHP micro- and nanostructures. In contrast to top–down strategies, these methods rely on in situ crystallization within confined environments, thereby enabling precise control over morphology while largely preserving the intrinsic optoelectronic properties of MHPs. Bottom–up synthetic approaches exploit capillary forces, wettability contrasts, and geometrical confinement in flexible or rigid templates such as PDMS, anodized aluminum oxide, or Si/SiO₂ microchannels to direct nucleation and crystal growth. These methods are particularly well suited for applications requiring highly ordered MHP with controlled geometry, such as microdisk lasers, photodetectors, integrated optical elements, and anisotropic charge- or exciton-transport platforms. Their main strength lies in the combination of spatial confinement and controlled crystallization kinetics, which enables the formation of high-quality, often single-crystalline structures with tunable width, thickness, spacing, and composition. On the other hand, the main limitations are related to scalability, reproducibility, and processing and may suffer from template deformation, defect formation, or undesired secondary growth if crystallization kinetics are not carefully controlled. Physical mask-assisted growth and lift-off methods offer better compatibility with lithographic patterning, multicolour integration, and device architectures, but they introduce additional processing steps and possible contamination or material degradation. Overall, while bottom–up approaches excel in preserving crystal quality and enabling low-damage fabrication, their successful implementation critically depends on precise control over nucleation and growth dynamics, with trade-offs between scalability, structural fidelity, and process complexity.

Applications of structured MHP Materials

Optoelectronic applications

Metal halide perovskites represent one of the most effective material platforms for optoelectronics, owing to their unique combination of a tunable bandgap, high absorption coefficient, high photoluminescence quantum yield, high charge-carrier mobility, and notable defect tolerance.⁸⁴ MHPs have been successfully integrated into a wide range of optoelectronic devices, enabling the fabrication of systems with competitive performance.^85–87 Micro- and nanostructuring MHP crystals and thin films can further enhance material performance while enabling new functionalities. In particular, micro- and nanostructuring represents a key step toward achieving compatibility between MHP-based devices and silicon photonics, while also opening opportunities for device miniaturization. This section focuses on the advantages of MHP micro- and nanostructuring demonstrated in recent studies on MHP-based photodetectors, light-emitting diodes, and color conversion systems.

Photodetector devices

Photodetectors are key optoelectronic devices with broad real-world applications, including optical communication systems, imaging and cameras, LiDAR, chemical and environmental sensing, healthcare monitoring, and security systems. Their performance is commonly described using several key figures of merit, the most representative being responsivity (R), detectivity (D*), time response, and linear dynamic range (LDR). Responsivity quantifies how efficiently incident photons are converted into an electrical signal and is defined as the generated photocurrent per unit of incident optical power. It is measured in A W⁻¹ and reflects the efficiency of photogeneration, charge transport, and charge collection at a given wavelength. Detectivity is closely related to responsivity but also accounts for noise, thus describing the ability of a device to detect weak optical signals. It is expressed in Jones and is particularly relevant for applications such as night-vision imaging and optical communications. The time response describes how quickly a photodetector reacts to changes in light intensity and depends on the carrier diffusion length and the charge extraction time at the electrodes. Fast response times, typically in the microsecond to nanosecond range, are essential for imaging, optical communications, and LiDAR. The linear dynamic range defines the range of incident light intensities over which the photodetector response remains linear. At low light intensities, device output is dominated by noise, which is strongly influenced by defect trap states and carrier mobility. At high light intensities, deviations from linearity and saturation typically occur due to charge recombination, limited charge extraction, thermal effects, and intrinsic absorption limits. A large LDR is technologically important, as it enables accurate measurements under fluctuating illumination, reliable operation in bright environments, and improved imaging quality. LDR is measured in decibels (dB), and commercially available silicon photodetectors typically exhibit values of approximately 120 dB. MHP-based photodetectors have, in recent years, achieved performance levels competitive with those of commercially available silicon photodetectors. Notably, micro- and nanostructuring of MHP materials can further enhance device efficiency and enable new functionalities. Compared to thin films, MHPs structured into one-dimensional (1D) crystals, such as microwires with high crystalline quality, promote directional charge transport and extend carrier lifetimes, reducing recombination. These improvements lead to more efficient photocarrier extraction, resulting in better device performance, including faster time response, higher responsivity, and enhanced detectivity.

Feng et al.⁸⁸ recently reported a photodetector based on MAPbBr₃ microwires, produced over a large area using the dynamic template-assisted coating approach (Fig. 8A(i)–(iii)). The photodetector exhibits a responsivity of 1660 A W⁻¹, a detectivity (D*) of 3.9 × 10¹⁴ Jones, a time response with rise time of t_r = 98.7 µs, decay time t_d = 56.6 µs, and a linear dynamic range (LDR) of 160.3 dB (Fig. 8A(iv)–(vi)). These enhancements are enabled by a combination of microwire crystal morphology and high crystal quality, achieved through optimized growth conditions and the use of methylamine trifluoroacetate as a passivating agent. The passivating agent also acts as a protective surface layer, providing enhanced hydrophobicity and improved stability, with nearly unchanged performance after 300 hours of testing under ambient conditions (85% RH, room temperature). This work demonstrates that a combination of high crystal quality and tailored morphology can produce devices capable of competing with current silicon technology.

Fig. 8. (A) (i) Large area MAPbBr₃ microwires on silicon and flexible polyimide substrates. (ii) PL image and (iii) planar and cross-sectional SEM images of the microwires synthesized by dynamic template-assisted coating. (iv)–(vi) Photodetectors performance, including: (iv) photocurrent and responsivity at different light intensities; (v) detectivity at different wavelengths and noise at different frequencies (inset); (vi) time response. Adapted with permission from ref. 88 Copyright 2025, Springer Nature. (B) (i) Schematic of a flexible multipixel photodetector array; (ii) photodetector array conformally attached to a walnut surface, resembling an eyeball shape and (iii) corresponding photocurrent map acquired using an H-shaped mask. Adapted with permission from ref. 89 Copyright 2021, Wiley-VCH. (C) (i) Conformal PPG sensors based on an MHP photodetector array and (ii) acquired pulse signal. Adapted with permission from ref. 90 Copyright 2024, Wiley-VCH.

Enhanced device performance has also been achieved by Hu et al. using MAPbI₃ and MAPbBr₃ microwire array-based photodetectors in a planar configuration.⁹¹ The arrays were fabricated using a PDMS template filled with precursor solution and heated in a closed system. MAPbI₃ devices exhibit high performance due to a reduced trap density compared to the corresponding polycrystalline thin films produced by conventional spin-coating methods. These devices achieve a responsivity of 118 A W⁻¹, detectivity up to 10¹³ Jones, time response of the order of hundreds of microseconds, and an LDR of 112 dB, and high operational stability, with no signal attenuation after 20 min under pulsed illumination. These results further confirm that 1D MHP structures can effectively enhance device performance.

Beyond MHP microwire arrays, also microplatelet arrays can enhance photodetector performance compared to thin films, for example in high-definition imaging devices. Compared to continuous films, photodetector pixel arrays can achieve improved spatial resolution, as each pixel operates independently, enabling parallel detection. Furthermore, independent pixel operation allows the whole device to maintain functionality under inhomogeneous illumination; for example, saturation may occur only in one or some of the illuminated pixels rather than across the entire detector. This reduces signal loss and preserves information across different light levels.

Li et al. demonstrated a MAPbBr₃ single-crystal microplatelets based photodetector array for real-time light trajectory tracking and visible-light imaging.⁹² The MAPbBr₃ single-crystal microplatelets, with typical diameters of 10–15 µm, were synthesized by confining the perovskite precursor solution in microholes defined in a PMMA thin film. The PMMA served both as a template and as an insulating layer to prevent short circuits between the top and bottom contacts in a vertical configuration. The final device consisted of a 3 × 3 pixel array and exhibited a low dark current (10⁻¹² A), along with a responsivity and detectivity under weak illumination (0.01 mW cm⁻² at 520 nm) of R = 0.38 A W⁻¹ and D* = 1.7 × 10¹² Jones, respectively. Furthermore, the presence of PMMA ensured enhanced stability, with only ≈10% reduction in photocurrent after up to 60 days of storage under ambient conditions. This configuration effectively enabled light trajectory tracking, with only the illuminated pixel showing an increase in photocurrent, while non-illuminated pixels remained unchanged.

In another study, Wu et al.⁸⁹ reported a 10 × 10 photodetector pixel array based on CsPbBr₃ microplatelets sandwiched between two flexible parylene C films acting as substrate and encapsulant (Fig. 8B(i)). The array was fabricated using a substrate with asymmetric wettability, which confined the precursor solution to specific hydrophilic regions. The resulting device exhibited a detectivity of 3.94 × 10¹² Jones at an illumination intensity of 4 µW cm⁻², with a corresponding responsivity of 3.15 A W⁻¹. CsPbBr₃ microplatelet arrays demonstrated high structural stability, as evidenced by unchanged XRD patterns, while the devices exhibited high operational stability with only a 5% reduction in photocurrent after 6 weeks of storage under ambient conditions. The array also demonstrated mechanical robustness, showing no performance degradation under up to 50% compressive strain or after repeated bending cycles. The potential application of this device as an artificial retina was demonstrated by testing it on a curved surface resembling an eyeball; only illuminated pixels produced significant photocurrent, while non-illuminated pixels showed negligible signals (Fig. 8B(ii) and (iii)).

In addition to this work, other reports indicate that photodetector arrays typically exhibit reduced mechanical rigidity and enhanced resistance to mechanical stress, opening opportunities for advanced flexible and conformal devices. In particular, Yin et al.⁹³ reported a method to fabricate MHP microwire arrays on flexible substrates by combining the use of a PDMS stamp with surfaces modification using protonated (3-aminopropyl)triethoxysilane. This surface treatment promotes the interaction with MHP precursors, allowing controlled nucleation and growth while improving mechanical stability through effective crystal anchoring. Furthermore, the microwire array exhibited high environmental stability, with no changes in the XRD pattern after up to 60 days under ambient conditions (25–30 °C and 50 ± 5% humidity). The resulting flexible microarray photodetectors showed promising performance, reaching a responsivity of 1.66 A W⁻¹ and a detectivity of 10¹² Jones, with performance preserved after bending.

The compatibility of perovskite arrays with wearable photodetectors has also been demonstrated by Xu et al.⁹⁰ with an array of CsPbBr₃ micropixels grown by CVD on flexible polyimide substrates. The growth approach is based on the patterning of the substrates with octadecyltrichlorosilane to obtain areas with different surface energies. This method enabled crystal growth only on non-functionalized areas, where compatibility with the substrate is higher. The resulting micropixel photodetector array yielded promising performance, including a responsivity and detectivity of approximately 48 A W⁻¹ and 10¹³ Jones, respectively, and was demonstrated suitable for photoplethysmography (PPG) wearable sensors with high sensitivity (Fig. 8C). The device also exhibited stable photocurrent after 12 h in a humid environment (≈70% humidity), while the PPG sensor enabled reliable pulse monitoring even after 12 h of continuous wear.

Another useful strategy to boost the performance of MHP-based photodetectors is to fabricate tailored geometries, such as gratings or metasurfaces, to increase the optical path length, achieve enhanced light absorption, and consequently obtain higher sensitivity and improved responsivity.

Micro- and nanostructuring can indeed help reduce optical losses, including both reflection and transmission losses, which typically affect MHPs due to the high refractive index contrast with air, as well as due to the reduced thickness of the active layer (Fig. 9A(i)).

Fig. 9. (A) (i) Flat MHP crystal affected by reflection losses and (ii) Moiré-structured MHP crystals under irradiation, with suppressed optical losses. (iii) Cross-sectional SEM image of Moiré-structured MHP crystals. (iv) and (v) Comparison of photocurrent and responsivity in MAPbBr₃-based photodetectors using flat, single-grating, and Moiré-structured crystals. Adapted with permission from ref. 94 Copyright 2022, Wiley-VCH. (B) (i) Schematic of structured MHP interaction with polarized light. (ii) SEM images of structured MHPs and (iii) photocurrent dependence on polarization angles.⁹⁶ Adapted with permission from ref. 96 Copyright 2023, Wiley-VCH. (C) (i) MHP microwire array device with bandgap gradient and (ii) bandgap tuning via anion exchange; (iii) photocurrent response under irradiation at different wavelengths along the wires, and iv) relative reconstructed monochromatic peaks across the 400–650 nm wavelength range.⁹⁷ Adapted with permission from ref. 97 Copyright 2023, Wiley-VCH.

With this aim, Li et al.⁹⁴ adopted a device configuration based on MAPbBr₃ crystals featuring a double-grating structure encapsulated between two PDMS templates. The bottom and top gratings were fabricated at a specific angle relative to each other to induce the formation of a Moiré lattice (Fig. 9A(ii) and (iii)). This structure ensures reflection suppression by the upper grating and waveguide-mode excitation in the high-index MHP, resulting in efficient interaction between light and the MHP active layer, thus enhancing light absorption and optoelectronic properties. The fabricated device exhibited superior performance compared to flat materials (Fig. 9A(iv) and (v)). The device with the Moiré lattice demonstrated a responsivity of R = 1026.5 A W⁻¹, detectivity of D* = 1.05 × 10¹⁴Jones, and a linear dynamic range of LDR = 103 dB. In contrast, the flat device showed much lower performance, with R = 157.2 A W⁻¹, D* = 1.62 × 10¹² Jones, and LDR = 62 dB. In addition, the highly hydrophobic PDMS layers provided effective protection against humidity and oxygen, resulting in significantly improved device stability. The device retained 95% of its initial photocurrent after 223 days under ambient conditions. Moreover, even after exposure to water vapor for 1 hour, the MHP preserved its crystallinity, as well as the photocurrent remained at 98.4% of its initial value.

Similar results were reported by Song et al.,⁹⁵ with perovskites structured with a top and bottom grating arranged to form a Moiré lattice, further corroborating the potential of this approach to enhance device performance. Also in this case, the device exhibited a good stability retaining nearly 80% of the initial photocurrent after 40 days of air in ambient conditions.

Microstructuring can provide MHP materials with new functionalities, including polarization sensitivity. Beyond measuring light intensity and wavelength, access to polarization information enables improved imaging under challenging conditions and enhanced security through unique optical fingerprints. Microwire structuring has enabled the demonstration of a full Stokes detector capable of sensing different light polarizations without metasurfaces or filters.⁹⁸ The device is based on (R/S-NEA)PbI₃ (NEA = naphthyl ethylamine) MHP microwires, structured using a PDMS template. In this system, the helical long-chain structure dictated by chiral NEA cations provides sensitivity to circularly polarized light, while the nanowire array structure ensures detection of linearly polarized light. By tuning the geometric parameters of the wires via the PDMS template, the device can precisely control polarization recognition, with maximum photocurrent when the polarization is parallel to the nanowire axis and minimum photocurrent when it is perpendicular, achieving a peak-to-valley ratio (PVR) of 1.5. In addition to R/S-NEA, other cations can also impart chirality to MHP structures. For instance, two-dimensional MHP nanowires based on chiral α-PEA (α-PEA = α-phenylethylamine), such as (R/S α-PEA)₂PbI₄, have demonstrated photodetection sensitivity to both circularly and linearly polarized light, arising from the chiral cations and the anisotropic wire morphology, respectively.⁹⁹ These nanowires, fabricated via a capillary bridge approach, exhibit high crystallinity and well-defined crystallographic orientation, leading to promising device performance, including a PVR of 1.6 with maximum photocurrent when the linearly polarized light is aligned parallel to the wire axis. In addition, the nanowires maintain their crystal quality after 30 days under ambient conditions, with no detectable changes in the XRD pattern. Furthermore, Bai et al. reported the synthesis of chiral 3D MHP microwires using (R/S)-1-(pyridine-4-yl)ethan-1-amine (R/S-PyEA), yielding (R/S-PyEA)Pb₂Br₆ with high crystal quality and strong crystallographic orientation.¹⁰⁰ As in previous studies, the cation chirality enables sensitivity to circularly polarized light, whereas the wire geometry determines the response to linearly polarized light. The resulting high-quality microwires enable full-Stokes photodetection with a PVR of 1.52 and excellent operational stability, with no performance degradation after up to 1060 switching cycles.

Similar polarization sensitivity has been observed in MAPbCl₃ single crystals structured with a grating using a space-confined growth approach, where the precursor solution is confined between the substrate and a patterned PDMS template (Fig. 9B(i) and (ii)).⁹⁶ The resulting device exhibits a PVR of 7.15, highlighting the enhanced sensitivity enabled by microstructuring (Fig. 9B(iii)). In addition, no significant changes in either the photocurrent or the dark current were observed after 45 days under ambient conditions, demonstrating the excellent stability of the structured MHP single crystals.

Finally, structured crystals can also offer advantages for the fabrication of color-cognitive devices, enabling a wide spectral response without the need for any filters and paving the way for compact and miniaturized spectrometers. MHP microwires with a bandgap gradient can provide a full-spectrum response in the visible-near IR range by inducing a gradual variation of the halogen composition along the wire (Fig. 9C(i) and (ii)).⁹⁷ The microwires with the bandgap gradient have been obtained using a post-synthesis anion exchange process based on the controlled dipping of the microwires into an octadecene-based solution of the specific halide salts. Interestingly, this approach enables control of the gradient formation by properly setting the anion exchange time. The as-prepared devices exhibit high performance when illuminated with a 405 nm laser under 5 V bias, including a responsivity of R ∼ 10³A W⁻¹, detectivities over 10¹⁵Jones, rise time of MAPbBr₃t_r = 266 µs, decay time t_d = 322 µs, and linear dynamic range LDR = 154 dB, under an irradiance of 120.5 mW cm⁻² and 5 V bias. The synthesized microwires have been integrated into an array of photodetectors, which demonstrated the ability to produce different response spectra when irradiated with wavelengths ranging from 400 to 760 nm in intervals of 14 nm, with the reconstructed spectra closely matching the ideal spectra obtained with commercial instruments (Fig. 9C(iii) and (iv)). In addition, the microwires show good environmental stability, with no detectable changes in the XRD pattern after 10 days under ambient conditions, while the device exhibits excellent operational stability, maintaining its performance after 200 hours of continuous irradiation at 405 nm.

Light emitting diodes and color conversion systems

Metal halide perovskites have attracted significant attention for lighting technologies thanks to their appealing optoelectronic properties, such as tunable emission wavelength, narrow spectra with high color purity, and high quantum yield. These features have enabled light-emitting diodes (LEDs) with excellent color purity and high luminescence efficiency, reaching EQE values above 20%, and offering the potential to meet the stringent requirements of ultra-high-definition displays.

As in the case of photodetectors, micro- and nanostructuring MHPs offer several advantages over continuous thin films, opening new opportunities for next-generation display technologies, where arrays with reduced pixel size and pitch are required. Miniaturizing MHPs to the micro- and nanoscale can enhance light-emitting performance by combining high luminance, fine resolution, and low energy consumption. High resolution is especially relevant for near-eye displays, which are emerging systems designed for use at very short viewing distances and essential for virtual, augmented, and mixed-reality devices, as well as smart glasses. Because these displays operate very close to the eye, extremely high resolution and pixel density are needed to avoid visible pixelation.

In this context, Zou et al.⁸³ reported a MHP-based LED array comprising 100 square pixels with 30 µm lateral dimensions, produced by a photolithographic dry lift-off method that prevents solvent exposure and preserves MHP quality. The vertical device structure consists of MHPs crystals sandwiched between a poly(3,4-ethylenedioxythiophene) polystyrenesulfonate (PEDOT:PSS) electrode and 2,2′,2″-(1,3,5-benzenetriyl)-tris(1-phenyl-1H-benzimidazole) (TBPi) electron transport layer, topped with a LiF/Al contact (Fig. 10A(i) and (ii)). The devices emit in the green region, with an electroluminescence peak at ∼519 nm and a linewidth of 19 nm. They show encouraging performance, with current and luminous efficiencies of 3.85 cd A⁻¹ and 1.86 lm W⁻¹, an EQE of 1.24%, and a maximum luminance of 13 043 cd m⁻² at 8.4 V (Fig. 10A(iii)).

Fig. 10. (A) (i) Device architecture of pixelated LED based on MHP crystal arrays; (ii) dark-field micrograph of the MHP crystal arrays; (iii) EQE and current-efficiency plots of the pixelated LED. Adapted with permission from ref. 83 Copyright 2020, American Chemical Society. (B) (i) and (ii) SEM images of nanocrystals formed at specific positions within wells on a silicon substrate; (iii) schematic and (iv) cross-sectional SEM image of nanocrystal-array LED with (v) corresponding electroluminescence spectra; (vi) and (vii) electroluminescence imaging of LED based on nanocrystal arrays. Adapted with permission from ref. 58 Copyright 2023, Springer Nature.

Further progress was reported by Jastrzebska-Perfect et al.,⁵⁸ with the demonstration of nano-LED arrays containing 600–2400 pixel units connected in parallel, based on CsPbBr₃ crystals with nanometer-scale dimensions positioned in well-defined locations on the substrate (Fig. 10B(i)). The arrays were fabricated using a SiO₂ topographical template characterized by droplet-shaped wells on ITO, where ∼200 nm crystals preferentially form near the rounded corners. The crystals are coated with a 5 nm PMMA layer to suppress short circuits and current leakage, followed by TPBi and LiF/Al as the electron-transport layer and top contact. The devices emit at ∼520 nm, with an average electroluminescence power of ∼0.6 fW at 10.5 V. This confirms the capability of nanometer-sized MHP crystals to operate as functional LED elements (Fig. 10B(iii) and (vi)).

As an alternative to direct electroluminescence from red, green, and blue (RGB) emitting LEDs, color conversion systems have recently emerged as a valid platform for full-color displays. In these systems, blue light generated by a blue LED is converted into red and green light using color conversion materials that absorb the blue radiation and emit at longer wavelengths. MHPs are highly promising materials due to their high PLQY, narrow full width at half maximum (FWHM), and high extinction coefficient, which support high color purity and efficient color conversion even at low thickness.¹⁰¹ In this context, pixelated MHP structures offer a promising route for color-conversion films targeting high-resolution full-color display applications. Zhang et al.¹⁰² developed a nanostructuring strategy that produces aligned MHP quantum wires within a porous alumina template. By adjusting the pore diameter or MHP composition, different emission colors can be obtained. For instance, green and blue wires arise by using CsPbBr₃ perovskite with diameters of 5.7 and 2.66 nm, respectively, while red emission is achieved using mixed-halide compositions containing iodide and bromide. Interestingly, the as-fabricated nanowires demonstrate high stability in ambient condition (25 °C temperature and 30% relative humidity) with nearly no degradation after 7 days. The wires embedded in the alumina matrix can be structured into pixels as small as 180 nm. These structures were integrated with a UVA microdisplay panel as color-conversion elements and were shown to be compatible with hemispherical transparent substrates resembling contact lenses, highlighting their promise for near-eye display technologies.

Photonic applications

Metal halide perovskites have recently emerged as a versatile platform for photonic applications because of their unique combination of high refractive index,^105,106 strong excitonic effects^6,107 and efficient light emission.^108–111

Unlike conventional semiconductors, MHP can be processed into a wide variety of micro- and nanostructures with relative ease,^112,113 enabling precise control over light propagation,^72,74 confinement,^114–116 and emission.⁶⁷ Such structures can act as optical resonators,^117–120 waveguides,^121–123 and metasurfaces,^124–127 supporting a rich set of photonic phenomena ranging from lasing^128–130 and nonlinear optics to strong light–matter coupling.^131–133

The ability to engineer MHPs into tailored geometries opens new opportunities for integrated photonics,^134,135 on-chip light sources,^18,25 and quantum technologies.^136,137

Among the different strategies for MHPs micro- and nanostructuring, the design of metasurfaces has proven particularly effective in tailoring light–matter interactions at the nanoscale.

Recent advances have exploited intrinsic photonic band-structure engineering to achieve compact and highly directional chiral light sources.^138,139 In particular, spin-valley-locked MHP metasurfaces have been realized by folding the Brillouin zone and coupling the MHP emission to bound states in the continuum (BIC), thereby routing photons with opposite helicities into distinct momentum valleys.¹²⁶ Synthetic valleys in the photonic band are achieved by controlling the growth of FAPbBr₃ MHP arranged in a square lattice of nanocylinders. Under optical pumping, photons emitted by the MHP couple to the high quality factor (Q) BIC mode and are selectively routed to opposite valleys according to their spin. With increasing pump power, the high-Q BIC acts as the cavity mode for lasing, giving rise to lasing in two circularly polarized BIC vortices.¹⁴⁰ This strategy simultaneously delivers high chiral purity, narrow beam divergence, and wide emission angles, setting a new paradigm for ultracompact circularly polarized emitters and MHP-based chiral lasers.

An alternative approach to engineering the band structure in the strong coupling regime consists of infiltrating the MHP precursors solution into the air holes of the backbone structure.^127,141 Dang et al. introduced a MHP-based metasurface platform capable of sustaining exciton–polaritons at room temperature with remarkable tunability.¹⁰³ The authors employed 2D layered MHPs structured into subwavelength lattices of nanopillars (Fig. 11A(i)), which act as photonic crystals. Thanks to the large exciton binding energy and strong oscillator strength of 2D MHPs, these nanostructures reach the strong coupling regime, exhibiting Rabi splittings of more than 200 meV. Beyond simply demonstrating robust polariton formation, the study emphasizes how the photonic lattice geometry directly controls the polariton band structure. By adjusting lattice symmetry and periodicity, the dispersion can be engineered into distinct regimes—ranging from parabolic to multi-valley profiles—thus offering a high degree of freedom in tailoring light–matter hybrid states (Fig. 11A(ii)).

Fig. 11. (A) (i) Scheme and SEM image of perovskite metasurface. (ii) Comparison between experimental results of the angle-resolved PL and numerical simulations of the angular-resolved absorption. Adapted with permission from ref. 103 Copyright 2020, American Chemical Society. (B) (i) Schematic of RT BIC polariton condensation from an air-hole CsPbBr₃ photonic crystal lattice. (ii) SEM image of the structure. (iii) Angle-resolved PL spectra below and above threshold. Adapted with permission from ref. 35. Copyright 2024, Springer Nature. (C) (i) SEM Image of TiO₂ grating on a CsPbBr₃ waveguide (left panel) and energy versus in-plane momentum PL of the waveguide (right panel). (ii) Real space map and corresponding PL spectra for three different incident pump fluences. Adapted with permission from ref. 63 Copyright 2024, Wiley-VCH. (D) (i) PL spectra emitted from two edges of the waveguide couplers at 0° (top panel) and 90° (lower panel) above the lasing threshold. (ii) Real-space PL of the perovskite beam splitter above the threshold. Normalized PL spectra of the perovskite beam splitter with different excitation configurations and propagation directions. Adapted with permission from ref. 104 Copyright 2024, Springer Nature.

As discussed in previous sections, single-crystal MHPs can be directly structured via FIB milling to produce clean, precise metasurfaces, enabling tailored engineering of the polariton dispersion.³⁵ Wu et al. demonstrated BIC polariton condensation in MHP photonic crystal lattice at room temperature (RT) (Fig. 11B(i)), by using FIB milling on high-quality CsPbBr₃ single-crystal microplatelets (Fig. 11B(ii)). The photonic crystal periodicity was engineered to support a BIC at the Γpoint of the lower-polariton dispersion for transverse magnetic (TM) polarization (Fig. 11B(iii), left panel). As the pump density increased above the threshold, the emission became dominated by a two-lobe pattern across the spectrum, indicating the formation of BIC polariton condensates (Fig. 11B(iii), right panel). The presence of a BIC polariton mode in the hybrid system offers a state with a theoretically infinite Q factor, enabling ultrahigh gain while minimizing losses in the mode dispersion, thereby promoting polariton accumulation and condensation.

In contrast to the fabrication of metasurfaces or geometrical photonic crystals on MHP, an alternative strategy exploited confined growth into specific geometries to precisely tailor the optical response of MHP-based devices.¹²⁰

In their study, Polimeno et al. demonstrated RT polariton condensation in CsPbBr₃ single-crystal microplatelets structured via a low-cost capillary-bridge synthesis.⁶³ The CsPbBr₃ waveguide supports propagating and counterpropagating polariton modes, arising from the strong coupling between photonic modes naturally confined within the crystal and the MHP exciton (Fig. 11C(i)). By tailoring the geometry of the single crystal from a waveguide to microplatelets, it was possible to achieve sharp edges and exceptionally smooth surfaces that suppress non-radiative scattering. This optimized morphology supported whispering gallery modes (WGMs) that strongly coupled with MHP excitons, giving rise to WGM-based polaritons. With increasing excitation fluence, three distinct polariton condensate modes emerged, each with its own condensation threshold (Fig. 11C(ii)).

The geometry of polycrystalline MHP microwires can also be precisely tailored through a combined solution-processing and vapor-phase conversion method.¹²² Wang et al. demonstrated that polycrystalline hybrid MHP microwires exhibit excellent optical waveguiding performance, with propagation losses as low as 0.30 dB µm⁻¹, remarkably comparable to single-crystal systems despite the presence of grain boundaries. Through controlled halide substitution and composition grading, they achieved wavelength-tunable waveguiding spanning the visible to near-infrared range, highlighting the intrinsic versatility of MHP waveguides. Together, these results establish MHP microwires as promising building blocks for tunable and reconfigurable photonic circuits, as evidenced by their integration into more complex structures.¹⁰⁴ Active multifunctional photonic devices were fabricated on monocrystalline CsPbBr₃ MHP thin films using a top–down etching approach with focused ion beams. The etched microwires exhibited high-quality microlasing, acting as compact light sources for integrated photonic circuits and enabling angle-dependent propagation between coupled MHP-microwire waveguides. Above the lasing threshold, the upper microwires emit coherent light, with signals detected at both terminals to compare input and output PL spectra. Real-space PL imaging of the waveguide couplers shows that the intensity and direction of lasing propagation can be directly tuned by the coupling angle (Fig. 11D(i)). Leveraging this strategy, multiple MHP-based active integrated photonic devices were realized for the first time, like a micro beam splitter that coherently divided lasing signals on two arms (Fig. 11D(ii)).

Building on this concept, Kedziora et al. presented a major advancement in integrated photonic circuitry by demonstrating a versatile, template-assisted approach for fabricating large-scale, predesigned MHP crystal waveguides.⁷² This technique allowed the fabrication of complex microstructures, including microwires, couplers, and splitters, determined solely by the geometry of the template. The authors successfully achieved non-resonant stimulation of an exciton-polariton condensate at RT, which results in bright polariton lasing emanating from the transverse interfaces and corners of the MHP microstructures. The results from the PL measurements, particularly the far-field PL patterns, underscore the ability to create and control highly coherent light sources within these predefined geometries.

The findings presented in this section provide a proof of concept for active integrated photonic devices based on MHP semiconductors, outlining a promising pathway toward their practical implementation in integrated optical chips.

Summary and outlook

In recent years, MHP micro- and nanostructures have experienced significant advances in both fabrication strategies and integration into optoelectronic and photonic devices. Advanced synthetic and processing approaches now enable the realization of a broad range of geometries, featuring reduced dimensions in the nano- and micrometer scale and precise placement on well-defined substrate locations, while also allowing such structures to extend over large areas. These structured MHP systems exhibit optical and electronic properties that are difficult to achieve in their non-structured counterparts, establishing them as relevant building blocks for next-generation optoelectronic and photonic devices.

Despite the significant progress achieved in the fabrication and device integration, several challenges remain to be addressed in order to realize the full potential of structured MHPs. Future research efforts should focus on further improving MHP structures resolution, enabling more complex geometries, and expanding accessible compositions, while preserving high crystal quality and enhancing stability under device operating conditions. To enable the transition of structured MHPs toward real-world applications beyond the laboratory scale, the following aspects need to be carefully considered and addressed.

Perspectives on advanced fabrication and synthesis of structured MHPs

Different fabrication and synthesis approaches have been developed in recent years to produce high-quality MHP materials and structures with increasingly high resolution. The advantages and limitations of each technique are rationalized and summarized in Table 1. The performance of post-growth and direct-growth techniques is schematically compared based on their scalability potential, achievable resolution, compatibility with perovskite materials, versatility in terms of achievable geometries, and applicability across different substrates and device architectures.

Table 1. Comparative table of different fabrication routes, highlighting the advantages and limitations of each method. Scalability, resolution, compatibility, and versatility are rationalised based on data reported in the available literature. Performance is visually represented using a qualitative rating scale indicated by colored dots, where a greater number of dots corresponds to stronger performance.

	Scalability	Resolution	Compatibility	Versatility
Post-growth approaches
Mask-mediated indirect approaches
E-beam lithography + RIE
Photolithography + RIE
Direct milling approaches
FIB milling
Laser writing
Scanning probe lithography
Micro-and nanoimprint lithography
Direct-growth approaches
Dewetting-assisted growth
Substrate with wettability contrast
Capillary bridge
Microfluidics-assisted growth
Soft templates
Rigid templates
Physical mask-assisted growth
E-beam lithography + lift off
Photolithography + lift off

Metal halide perovskite structures with feature sizes ranging from the micrometer down to the nanometer scale can be effectively fabricated using top–down approaches such as EBL and photolithography. These techniques enable the preparation of resist masks on top of MHP layers, through which the material can be directly etched using dry etching processes. Despite significant progress to preserve structural material integrity have been achieved through careful design of both the resist mask definition and the etching protocols, the soft nature of MHPs remain a critical challenge. Solvent exposure and etching process can induce material degradation, leading to structures with rough edges and compromised optical properties. In addition, the use of alternative solvents and milder processing conditions compatible with MHPs often limits the resolution, preventing access to the state-of-the-art feature sizes achievable with conventional high-performance resists. Beyond these techniques, direct approaches such as focused ion beam milling, femtosecond laser ablation, and scanning probe lithography offer the advantage of avoiding the use of solvents while enabling high-resolution micro- and nanostructuring. However, despite their potential, direct techniques usually leave material residues that must be removed through post-processing steps. In this context, further advantages could arise from the use of thermal scanning probe lithography, which preserves the benefits of a solvent-free approach while potentially achieving high resolution and avoiding residues and additional post-processing. Thermal scanning probe lithography could indeed enable direct material modification by locally inducing decomposition or recrystallization, allowing precise removal or transformation of the perovskite.

In addition to conventional lithography approaches, nanoimprint lithography emerges as a promising top–down nanostructuring strategy for MHPs, combining nanoscale precision with an intrinsically non-destructive character. NIL techniques are poised to play a central role in the deterministic nanostructuring of MHP materials, effectively bridging the gap between nanoscale materials engineering and scalable device fabrication. Continued advances in mold design, process control, and imprint-assisted crystallization are expected to further improve pattern fidelity and phase stability across a broad range of MHP compositions. In particular, the convergence of NIL with additive strategies, hierarchical self-assembly, and dry transfer approaches opens new opportunities for damage-free nanostructuring of single-crystal and heterostructured architectures. As these techniques mature, NIL is likely to became a versatile platform for the large-area, low-cost manufacturing of MHP-based photonic and optoelectronic devices with precisely engineered functionalities.

As an alternative to top–down approaches, bottom–up direct growth of MHPs via liquid-phase processes has emerged as an effective route to micro- nanostructures formation, particularly through techniques such as microfluidic growth and dewetting-assisted growth. These approaches enable exploitation of the intrinsic properties of these materials, which are often strongly governed by crystallinity, dimensionality, and interfacial quality. By combining precursor-solution confinement with kinetic and thermodynamic control, these methods allow fine regulation of nucleation pathways and crystal orientation. In this context, interfacial engineering through chemical functionalization or surface-energy modulation emerges as a powerful strategy to direct site-selective crystallization.

As a result, direct-growth solution-phase strategies have enabled the realization of functional micro- and nanostructured MHP architectures that preserve structural integrity and physicochemical properties while providing a versatile and accessible platform readily integrable into advanced devices for both fundamental studies and next-generation optoelectronic and photonic applications. A wide range of geometries can be realized spanning different shapes and dimensions, including wires, square pixels, and circular microstructures. Moreover, these approaches enable access to dimensional regimes that are difficult to achieve using conventional techniques, such as quasi-1D or quasi-0D structures, where quantum confinement effects, emission anisotropy, and polarization control become dominant. These capabilities open new opportunities for directional emitters, integrated waveguides, and on-chip coherent light sources.

Beyond their intrinsic flexibility in achieving different sizes and shapes, these techniques provide precise control over crystal position and orientation on the substrate, which represent key requirements for their integration into final devices. Indeed, predefining crystal dimensions and placement through templates or substrate patterning enables rational pre-design of geometries compatible with device architectures. This approach could potentially enable semi-automated device fabrication using predefined masks, ensuring high reproducibility and rapid fabrication of complete devices, without the continuous process readjustment typically required for transferred or randomly grown crystals.

Despite their strong potential, several challenges remain, including the realization of even more versatile geometric features and the growth of single crystals with extremely high aspect ratios, characterized by widths on the order of hundreds of micrometers and thicknesses down to tens of nanometers. Furthermore, while these methods have been widely exploited for 2D and 3D MHPs, their effectiveness has not been demonstrated for quasi-2D MHPs, probably due to the complex crystallization equilibrium required to achieve phase-pure crystals. This limits the library of possible MHP compositions achievable with these strategies, precluding the exploitation of the potential of quasi-2D MHP, which exhibit features not achievable in their pure 2D or 3D counterparts. In parallel, the development of sustainable processing strategies based on green solvents and room-temperature conditions is gaining increasing relevance for scalable and environmentally responsible fabrication.

Ultimately, a critical frontier consists in the transition from proof-of-concept demonstrations to scalable manufacturing processes, where direct-growth approaches can be combined with printing techniques such as inkjet and roll-to-roll for large-area fabrication. This combination makes direct growth liquid-phase processes particularly attractive for real-world applications in photonics, lighting, and sensing technologies.

Perspectives on optoelectronic applications of structured MHPs

The increasing number of high-performance optoelectronic devices based on structured MHPs highlights their relevance as an emerging semiconductor platform. Micro- and nanostructuring MHP crystals and thin films plays a key role in improving not only device performance but also in providing new functionalities, enabling the development of compact and innovative optoelectronic devices.

Photodetectors based on structured MHPs have reached performance that are competitive with commercial silicon devices, combining high responsivity and detectivity with fast response times and wide linear dynamic ranges. These advances can be attributed to improved charge transport, reduced recombination, and enhanced spatial resolution achieved through micro- and nanostructures such as microwires, microplatelets, and pixel arrays. Furthermore, tailored geometries, such as gratings and Moiré lattices, enhance light absorption, while chiral microwires and structures with variable bandgap enable additional functionalities, such as polarization sensitivity and full-spectrum response. The compatibility of these structured MHPs with flexible substrates opens opportunities for wearable and conformal photodetectors.

In parallel, structured MHP LEDs benefit from improved spatial control and preserved crystal quality, enabling high resolution and high efficiency which represent important requirements for advanced display technologies, including near-eye and augmented-reality systems. Micro- and nanostructuring allows precise pixel definition and the resulting pixelated MHP structures can also function as color-conversion elements for full-color displays that are also compatible with compact and curved substrates.

Overall, MHP micro- and nanostructuring enables device miniaturization, supporting high-density photodetector and LED arrays with extremely small pixel sizes for next-generation imaging, sensing, and display applications. Miniaturized MHP arrays combine spatial resolution with multifunctional operation such as the simultaneous detection of light intensity, wavelength, and polarization, with potential impact in environmental monitoring, optical communication, and healthcare diagnostics.

However, several challenges must be addressed to fully realize this potential. Photodetector architectures have yet to be optimized to suppress crosstalk between adjacent pixels, an issue extensively addressed in silicon-based technologies but still largely unexplored for MHP systems. Furthermore, limited attention has been devoted to single-pixel performance and to pixel-to-pixel variability, which poses the risk of non-uniform performance across the detector array and requires complex calibration procedures. Parameters such as photo response non-uniformity (PRNU), which quantifies pixel-to-pixel variations in the photoresponse under uniform illumination arising from small differences in pixel dimensions or sensitivity, and dark signal non-uniformity (DSNU), which measures variations in the dark current and may reflect differences in defect density among pixels, should also be reported. The characterization of these parameters is essential to assess the uniformity of multipixel photodetectors and to evaluate their potential for practical applications beyond proof-of-concept demonstrations in which only a few pixels are investigated. For LEDs, current device architectures are often simplified, in part due to the tendency toward short circuits between top and bottom electrodes, which limits the use of multiple charge-transport layers (CTLs) typically required in MHP LED architectures. CTLs are essential components because they reduce the charge-injection barriers between the electrodes and the MHP, provide selective contacts for electrons and holes, facilitate charge-carrier transport, and help balance electron and hole injection, thereby suppressing charge accumulation and non-radiative recombination at the interfaces.¹⁴² To further optimize charge injection and transport, the creation of an energy-level cascade through the appropriate modification or sequential deposition of multilayer CTLs, for example PEDOT:PSS/poly[N,N′-bis(4-butylphenyl)-N,N′-bis(phenyl)benzidine] (Poly-TPD)/poly(9-vinylcarbazole) (PVK), could provide improved control over the electrode interfaces, leading to enhanced emission, higher luminance, and narrower linewidths in structured MHP-based LEDs.

In addition, the soft nature of MHP and their sensitivity to degradation require highly effective encapsulation strategies to enable their widespread use. Beyond laboratory-scale, encapsulation must ensure long-term stability under harsh operating conditions, including humidity, thermal cycling, mechanical stress, and prolonged illumination. For applications involving direct or prolonged interaction with the human body, such as wearable or healthcare devices, encapsulation materials must be biocompatible and fully impermeable to prevent leakage and ensure safety.

Perspectives on photonic applications of structured MHPs

The results discussed above clearly position MHPs as more than an efficient emissive material: they constitute a realistic platform for active and nonlinear photonic circuitry, and in particular for polaritonic integrated devices, where optical nonlinearities exceed by orders of magnitude those available in conventional χ(2)or χ(3)materials. This point is not merely conceptual. The combination of large exciton binding energies, strong oscillator strengths and straightforward micro- and nanofabrication has already enabled the realization of polaritonic waveguides, microcavities and interferometric elements operating at room temperature, with characteristic device lengths of only a few tens of microns working in this highly nonlinear regime.

A particularly relevant recent development is the observation of polariton lasing with net optical gain and nonlinear amplification in fully integrated waveguide geometries, over propagation distances of a few tens of µm. This regime, recently demonstrated in MHP polaritonic circuits, shows that losses—certainly higher than in silicon or silica platforms—can be compensated locally by exploiting bosonic stimulation and condensate-assisted gain.¹⁴³ From a device perspective, this dramatically changes the scaling problem of nonlinear photonics: functions such as optical switching, modulation, amplification or logic operations no longer require long interaction lengths or resonant enhancement over large footprints. In this framework, interferometric architectures—Mach–Zehnder-like geometries, directional couplers or compact splitters—with longitudinal sizes of a few tens of microns become realistic candidates for scalable nonlinear photonic circuits, where phase and intensity can be controlled optically and, in principle, electrically.

Low-threshold lasing represents another of the most tangible and interesting outcomes of MHPs and, in particular, MHP polaritonics. By relying on bosonic final-state stimulation rather than population inversion, MHP-based polariton lasers have demonstrated remarkably low thresholds, in some cases approaching or even surpassing those of conventional solid-state lasers.¹²⁹ These results are particularly appealing in the context of integrated photonics, as they point to coherent light sources that are compact, energetically efficient and naturally compatible with on-chip architectures, where low cost and power density are critical constraints. At the same time, a careful and critical assessment is required. The interpretation of ultra-low lasing thresholds hinges on an accurate determination of the effective polariton and photon densities involved in the condensation process. Effects such as exciton and polariton funneling, spatial inhomogeneities of the pump, and defect-assisted localization can strongly reduce the effective mode volume and artificially enhance the apparent gain. If not properly accounted for, these mechanisms may lead to an overestimation of device performance and blur the distinction between genuine polariton condensation and more conventional defect- or cavity-assisted lasing. Further studies, including independent repetitions of these experiments by different groups, are therefore needed to firmly establish the relevant density scales and to consolidate a reliable benchmark for low-threshold polariton lasing in MHP-based systems.

Beyond classical signal processing, these same ingredients naturally connect MHP polariton platforms to neuromorphic and non-von-Neumann computing paradigms. Exciton–polaritons intrinsically provide strong nonlinear responses, threshold behaviour, gain saturation and relaxation dynamics on ultrafast timescales, which are precisely the physical mechanisms required to emulate artificial neurons. Polariton-based neuromorphic concepts—ranging from reservoir computing to spiking neurons and binarized neural networks—have already shown clear advantages over linear classifiers and conventional photonic.^144–146 What makes MHPs particularly compelling in this context is not novelty per se, but practicality: low-cost fabrication, compatibility with predefined geometries, and operation across a broad spectral range from the blue to the near infrared. The recent demonstration of room-temperature polariton neural networks in perovskite waveguides¹⁴⁷ represents the first step toward hardware neuromorphic systems that do not rely on cryogenic platforms or complex epitaxial growth.

At the same time, it is important to be explicit about the open problems. For MHPs, material stability remains the most serious bottleneck, especially under continuous or high-power optical excitation. Photo-induced degradation, ion migration and chemical instability are not marginal issues; they directly affect device reproducibility and long-term operation. Addressing them will require more than incremental improvements, including robust chemical passivation, encapsulation strategies, surface functionalization and careful thermal management. Without these advances, the transition from proof-of-concept devices to reliable integrated circuits will remain limited.

Finally, the most critical step toward realistic applications is electrical injection. While optically pumped MHP polariton devices have now reached a high level of maturity, electrically driven architectures are still in their infancy. First demonstrations of electrically pumped MHP polariton lasers indicate that this route is viable, albeit currently at low temperatures and with complex device engineering.^148,149 Achieving stable, room-temperature, electrically injected polaritonic circuits would represent a genuine turning point, enabling fully integrated photonic and neuromorphic chips compatible with electronic platforms.

Conflicts of interest

The authors declare no competing financial interest.

Biography

From left to right: Ilenia Viola, Daniele Sanvitto, Rosanna Mastria, Luisa De Marco, Francesco Todisco, Aurora Rizzo, Francesco Saverio Difeo, Laura Polimeno.

The Advanced Materials, Devices and Photonics groups at CNR-NANOTEC focus on the development of innovative nanoscale-engineered materials and multifunctional devices for energy production, storage, and energy saving, together with the study of polariton quantum fluids and frontier light–matter interactions. Using advanced bottom–up and top–down fabrication approaches, we design and manufacture devices based on organic, inorganic, and hybrid materials on both rigid and flexible substrates, achieving functionalities beyond the current state of the art and at high Technology Readiness Levels (TRLs). Our research in nanotechnology and material science enables the development of eco-friendly hybrid materials and their integration into low-energy micro- and nanostructured optoelectronic devices. Current activities include perovskite solar cells, sustainable batteries, wearable electronics, multifunctional energy-saving devices, and advanced optoelectronic and photonic systems. On a more fundamental level, we investigate collective coherent phenomena in Bose–Einstein condensates in semiconductor materials, exploiting quasiparticles generated through the strong coupling between light and excited states of matter. Our studies address quantum fluids in superfluid and supersonic regimes, including vortex dynamics, non-equilibrium condensates, quantum-state correlations, and highly confined plasmon-polariton fields, while also exploring concepts for all-optical transistors, switches, and logic gates.

Data availability

No primary research results, software or code have been included and no new data were generated or analysed as part of this review.