Porous-lift-off germanium membranes as lightweight, flexible, and cost-effective template for III-V space solar cells: a review Authors: Azmat Ali, Radouane En-Nadir, Chahinaz Khouloud Mahboub, Imad-Eddine Mokeddem, Ahmed Ayari, Brieuc Mével, Jérôme Ripa, Artur Turala, Alexandre Chapotot, Priyanka Proost, Rajiv Sharma, Jinyoun Cho, Kristof Dessein, Abderraouf Boucherif Published: 2026 ABSTRACT The growing demand for low-cost, lightweight, and high-efficiency space technologies has driven the search for photovoltaic solutions which minimize material use without sacrificing performance. III-V multijunction solar cells remain the state of the art for space power applications due to their exceptional efficiency and radiation resistance. Among III-V solar cells architectures, Germanium (Ge)-based structure remains widely used; however, their broader deployment is limited by reliance on thick and expensive Ge substrates. Germanium membranes, however, offer a promising path to address these limitations through substrate reuse strategies. This review highlights progress in the development of Ge membrane as a transformative platform for space photovoltaics, with a focus on porous Ge-based lift-off approaches while reusing Ge substrate. These methods enable the fabrication of detachable, monocrystalline Ge membrane through pore etching, epitaxial growth, porous transformation, layer lift-off and substrate reuse, significantly reducing Ge consumption and device weight. Two main fabrication routes for porous Ge formation are discussed: lithography-based dry etching using Ge-On-Nothing approach and electrochemical etching. The fundamentals of pore formation, thermal reorganization of porous, and subsequent III-V epitaxy are reviewed. Key challenges related to process uniformity, detachment scalability, and industrial integration are examined, and potential strategies for scalable manufacturing are outlined. Introduction As the space industry grows, satellites play a vital role, ranging from observing the sun (Parker solar probe)1, exploring distant galaxies (Hubble space telescope)2 and studying planets (Mars reconnaissance orbiter, Juno)3 to enabling connectivity (Starlink)4. To power these satellites, the demand for efficient, reliable and stable solar cells has increased more than ever before. Among the various solar cells technologies, III-V multi-junction solar cells (MJSCs) are considered the gold standard for powering satellites owing to their unmatched efficiency, radiation resistance and durability5. However, their high production costs remain a challenge. Although for many years the primary research was focused on increasing the efficiency of space solar cells, the recent drastic reduction in launch costs, driven by reusable rockets such as SpaceX’s Falcon 96, together with launch capacity, has changed the landscape. This is exemplified by large-scale constellation such as Starlink, representing an extreme case where short operational lifetimes and reduced radiation exposure lead to the use of silicon solar cells, the only technology currently available at the required manufacturing scale. While silicon solar cells currently dominate large-scale low Earth orbit (LEO) constellations due to their scalability, this advantage is closely tied to short mission lifetimes and reduced radiation exposure. Additionally, the increasing number of satellites and associated concerns, such as space debris, are driving interest toward longer operational lifetimes and improved system efficiency7,8. In this context, III–V-based solar cells remain highly competitive due to their superior efficiency and radiation hardness, which can enable reduced array size, lower mass, and extended mission lifetimes at the system level, despite their higher material cost. Consequently, system-level cost, particularly photovoltaic assembly and solar wing manufacturing, has become dominating factor. Therefore, although lower launch costs, increased lunch capacity and reduced solar wing costs lessen the strict sensitivity to payload weight, lighter photovoltaics assemblies still provide important advantages by enabling larger deployable area, easier integration and improved mission flexibility. At the same time, the rapid growth of the space industry increased the demand for more affordable, lightweight and high-performance solar cells9,10. As a result, research communities are now focusing on strategies to lower solar cell costs by minimizing weight, increasing throughput and availability (e.g., reducing epitaxy growth time), reducing solar cell complexity and optimizing material use11–17. This shift has made cost-effective and low-mass photovoltaic technologies a major research priority. The first solar-powered satellite, Vanguard 1 launched into space in 1958 was based on single-crystal Si-based SCs with power conversion efficiency of 10%5. However, their heavy weight (with a thickness > 100 μm and a specific power of ≈0.38 W/g)18, combined with their proneness to degradation when exposed to high-energy protons, electrons and cosmic rays due to radiation-induced displacement in the material lattice result in a decrease in the lifetime of photogenerated carriers and limited long-term performance in space19,20. Despite these limitations, the recent surge of Starlink21 megaconstellation has resulted in the widespread deployment of Si solar arrays5. Currently, Si-based SCs with an efficiency of 16.9% under AM0 in space condition, manufactured by AzurSpace, are qualified for space mission18. Similarly, Copper indium selenide chalcopyrite-structure alloys with gallium (CIGS) thin-film solar cells are attractive due to their high radiation tolerance, light weight, and the potential to be manufactured on flexible substrates (e.g., polyimide) using low-cost techniques. However, they are not commonly used as a power source for satellites due to their low efficiency (about 16%, under 1 sun AM0 space conditions, though with impressive 1.96 W/g)22 and power losses caused by their low emissivity (0.18-0.3) in the IR region23. In contrast, solar cells based on III-V compounds semiconductor provide superior performance in terms of both efficiency and radiation hardness, making them preferred choice superior for high performance space applications24. Their higher End-Of-Life (EOL) efficiency allows for significantly smaller solar array areas compared to silicon-based counterpart, typically by a factor of 2–3 (i.e., roughly 50–65% smaller)25, which directly reduces size and mass of the spacecraft. Although silicon cells are less expensive per unit area at the bare cell level, their lower efficiency necessitates larger panel areas. This leads to increased system-level mass and cost due to numerous additional components, such as the panel substrate, face sheet (and adhesive), hinges, insulators, wiring etc., and associated integration effort26. When these system-level considerations, including manufacturing, assembly, and testing are taken into account along with low EOL efficiency, the apparent cost advantage of silicon is substantially reduced, further highlighting the advantages of III–V-based solar technologies for space applications. Since the first launch of GaAs-based powered satellites in 1977, space-based III-V MJSCs technology has seen remarkable progress in efficiency, cost reduction, thickness reduction, lightweight design, flexibility, high power-to-mass ratio and radiation resistance19. Starting with single junction gallium arsenide (GaAs) with efficiency of about 22% with the thickness >50 mm in the 1970s27, the field evolved with the introduction of multijunction architectures such as GaAs/Ge28 and GaInP/GaAs/Ge29, reaching efficiencies of 32–38% in the 2000s and early 2010s29–31. However, the more efficient to our knowledge, reported, commercially available solar cell device (QJ 4G32-Advanced (8 × 8 cm2)) based on AlInGaP/AlInGaAs/InGaAs/Ge32 MJSCs achieves an end-of-life (EOL) efficiency of 29.30% under (AM0, 1353 W/m2) at a 1 MeV electron irradiation fluence of 5 × 1014/cm2 (more than 10-yr LEO), and 28.3% under (AM0, 1353 W/m2) at a 1 MeV electron irradiation fluence of 1 × 1015/cm2 (15-yr GEO) as reported by AZUR SPACE Solar Power GmbH18. Similarly, Inverted metamorphic (IMM - ɑ)33 n-on-p solar cell grown on either Ge or GaAs substrate, achieves an EOL efficiency of 29.12% under (AM0, 1353 W/m2) at a 1 MeV electron irradiation fluence of 5 × 1014/cm2 (more than 10-yr LEO), and 27.84% under (AM0, 1353 W/m2) at a 1 MeV electron irradiation fluence of 1 × 1015/cm2 (15-yr GEO), as reported by Rocket Lab34. In comparison, the IMM4J Gen II device, an experimental demonstration35, offers a significantly higher power-to-weight ratio (~ 2.5 W/g) with an EOL efficiency of ~27.5% after 1 MeV electron irradiation at a fluence of 2.5 × 1014 cm−2 under (AM0 1367 W/m2). When comparing this to commercially available devices mentioned above from AZUR SPACE Solar Power GmbH and Rocket Lab, their EOL efficiencies approaching ~29% under comparable or higher irradiation conditions, however with lower P/W ratios (~ 0.5 W/g). Triple-junction (3j) upright lattice matched III-V MJSCs are heavy (with a specific power of 0.4–0.8 W/g)5,36 are grown on thick germanium (Ge) substrates using metal-organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE) for precise control down to a monolayer37–40. While the Ge substrate serves both as a growth platform and as a bottom subcell, its scarcity makes it a critical raw material41, and the use of thick Ge substrates contributes significantly to the device cost, accounting for 30–40% of the total MJSC device cost16,42. Additionally, Ge substrate’s thickness contributes significantly to the device’s weight43, thereby affecting power-to-mass ratio (P/M), a critical metric for space photovoltaics. It has been reported that effective performance requirements for space solar generators are typically around power-to-volume (P/V) = 60 kW/m3, P/M = 200 W/kg, and a total power generation capacity of approximately 150 kW5. Thus, using a thin Ge substrate is necessary, as it will not only reduce the costs but also address both Ge scarcity and weight of the device. In this regards, detachable single-crystalline Ge membranes with the thickness (few microns) provide a promising solution for reducing Ge consumption and enhancing device cost-effectiveness44. The fabrication of Ge membranes involves the sequential process of producing an epi-ready Ge growth template for III-V layer, either by porosification and subsequent epitaxial Ge growth or porosification and thermal annealing to form an engineered Ge substrate45. This Ge membranes serves as templates for III-V epitaxial growth and subsequent device processing46. Additionally, the membranes can be transferred to host substrates46, significantly reducing material use, and allow for the reuse of expensive Ge substrate16. For instance, a thinned Ge subcell in a GaInP/Ga(In)As/Ge triple-junction solar cell, at just 5–10 μm, reduces subcell weight by 90% while retaining over 95% of its current43. Additionally, heat absorption beyond 1600 nm is reduced by more than 85%, further improving the device functionality43. Regarding the manufacturing of detachable membranes and subsequent reuse of the substrate, numerous strategies have been developed over the years, as shown in Fig. 1. These includes the epitaxial lift-off via selective chemical etching13,47, controlled spalling for strain-induced layer separation17,48, Smart CutTM, which utilizes ion implantation and wafer bonding for layer transfer49,50, Laser Lift-off, which relies on laser-induced decomposition or ablation at an engineered interface to separate epitaxial layer51,52, Remote epitaxy53–55, which enables layers to inherit the crystallographic properties of their underlying substrates while remaining physically detachable. Epitaxial layer TRANsfer (ELTRAN®), which combines wafer bonding with porous silicon etching and hydrogen annealing to enable layer separation56,57.Fig. 1Schematic representation of various techniques used for the fabrication of Ge and/or GaAs nanomembrane for III-V device applications.The images shown in the left panel are reproduced or adapted from the following sources: Smart Cut™ technology (copyright© Soitec); epitaxial lift-off (ELO) (reproduced from13 with permission from Springer Nature B.V., permission conveyed through Copyright Clearance Center, Inc.); laser lift-off (reproduced from51 with permission from Springer Nature B.V., permission conveyed through Copyright Clearance Center, Inc.); remote epitaxy (reproduced from71 with permission from Springer Nature B.V., permission conveyed through Copyright Clearance Center, Inc); controlled spalling (reproduced from 48 licensed under the CCBY 3.0); and porous lift-off (reproduced from80 licenced under the CCBY 4.0). The right panel illustrates the three porous lift-off approaches: GeON, Biporous mesoporous Ge, and single mesoporous Ge layer + Ge-epi, adopted from45 with permission from IEEE. Schematic representation of various techniques used for the fabrication of Ge and/or GaAs nanomembrane for III-V device applications. The images shown in the left panel are reproduced or adapted from the following sources: Smart Cut™ technology (copyright© Soitec); epitaxial lift-off (ELO) (reproduced from13 with permission from Springer Nature B.V., permission conveyed through Copyright Clearance Center, Inc.); laser lift-off (reproduced from51 with permission from Springer Nature B.V., permission conveyed through Copyright Clearance Center, Inc.); remote epitaxy (reproduced from71 with permission from Springer Nature B.V., permission conveyed through Copyright Clearance Center, Inc); controlled spalling (reproduced from 48 licensed under the CCBY 3.0); and porous lift-off (reproduced from80 licenced under the CCBY 4.0). The right panel illustrates the three porous lift-off approaches: GeON, Biporous mesoporous Ge, and single mesoporous Ge layer + Ge-epi, adopted from45 with permission from IEEE. In epitaxial lift-off, a sacrificial layer (typically AlGaAs) is inserted between the substrate and the epitaxial device layers and selectively removed using chemical etchants such as HF, enabling the release of the active film. While this approach can produce high-quality freestanding layers, it is fundamentally limited by slow lateral etching kinetics, often resulting in processing times on the order of several hours for full wafers13,58. In addition, the substrate surface is often degraded after lift-off, requiring chemical–mechanical polishing (CMP) for reuse, which increases cost and reduces substrate thickness58,59. It should be noted that the use of HF-based chemistries limits its compatibility with Ge substrates due to parasitic etching of Ge. Controlled spalling offers an alternative approach based on stress-induced fracture, where a tensile stressor layer (e.g., Ni) is deposited to initiate crack propagation at a controlled depth within the substrate. This method enables relatively fast and large-area layer transfer and has been demonstrated for both lattice-matched and metamorphic III–V structures17. However, challenges remain in achieving uniform thickness control and minimizing surface damage. In particular, spalled Ge surfaces often exhibit morphological defects such as arrest lines and surface roughness, which necessitate post-processing (e.g., CMP) prior to subsequent epitaxial growth60–62. Smart CutTM technology, originally developed for silicon-on-insulator (SOI) fabrication, combines ion implantation and wafer bonding to achieve precise layer transfer49,50. The thickness of the transferred layer is defined by the implantation depth, offering excellent control and scalability, and the technology is already widely commercialized for Si-based platforms such as Photonic-SOI, Imager-SOI and Connect RF-GaN, etc. by Soitec63. Its extension to germanium-on-insulator (GeOI) has also been demonstrated64; however, ion implantation in Ge introduces defects such as vacancy clusters and nanocavities, which can degrade material quality and device performance65. In addition, surface roughness after layer splitting requires CMP, and interface passivation remains a critical challenge, particularly for ultra-thin Ge layers66. While Smart CutTM method has proven versatile for Ge integration, enabling NIR detectors on Ge-on-Glass (GOG) with performance parity to bulk Ge67, and the creation of Ge-on-Sapphire (GeOS) templates68, there is limited literature though exists to explore the feasibility of this specific process for high-efficiency solar cells69. Remote epitaxy has recently emerged as a novel technique in which interaction between the substrate (e.g., GaAs, InP, and GaP) and the epilayer occurs through a 2D material (e.g., graphene), which is sufficiently thin and electrically penetrable to guide the epitaxial orientation of the overlayer, allowing the epitaxial film to inherit the crystallographic orientation of the substrate while remaining weakly bonded for subsequent exfoliation53,54,70,71. This approach has enabled the demonstration of flexible III–V devices and substrate reuse without conventional sacrificial layers72. However, the method is still in its early stages of development and faces challenges related to material quality, defect control, and scalability53. Moreover, its direct applicability to non-polar substrates such as Ge is limited, requiring modified approaches such as anchor point nucleation (APN) to facilitate epitaxial alignment55. Laser lift-off, initially developed for GaN systems, relies on localized laser-induced decomposition or ablation at an engineered interface to separate the epitaxial layer52. Extensions of this approach to III–V systems (e.g., GaAs) using absorbing sacrificial layers (InGaAsN) have been demonstrated to separate a GaAs membrane from its GaAs substrate and transfer it onto a polymer substrate51. While laser lift-off offers rapid and localized layer transfer, it is currently limited by small-area processing, potential thermal or structural damage73, and a lack of demonstrated high-efficiency device integration, particularly for photovoltaic applications. The ELTRAN® process, pioneered for silicon, utilizes a porous sacrificial layer combined with wafer bonding and thermal treatment to enable layer transfer. This method provides a pathway for substrate reuse and high-quality epitaxial growth but involves multiple high-temperature and multi-step processing stages57. ELTRAN®-based approaches have been explored for the fabrication of thin crystalline silicon layers for solar cell applications, also known as Direct Gas to WaferTM technology74,75. Using the concept of ELTRAN® technology, NexWafe76, a company spun out from Fraunhofer Institute for Solar Energy Systems ISE in 2015, has developed EpiNex® technology, enabling the production of lightweight, cost-effective and ultra-thin (70 µm) silicon solar cells, for powering low earth orbit satellites. The key aspects of this technology involve the anodic etching of Si substrate to form porous layer, followed by high-temperature annealing (1120 °C) to reorganize the structure into a high porosity detachment layer at bottom part and a closed surface on top acting as template for the subsequent epitaxial growth using trichlorosilane (SiHCl3) as a precursor in hydrogen (H2) atmosphere74,75. A final separation or lift-off steps enables the detachment of epitaxial wafer from the substrate and reuse of the substrate. The porous layer plays a dual role as both a detachment interface and a template for epitaxy, forming the conceptual basis for analogous approaches in other material systems such as porous lift-off for Ge, which will be discussed below. Apart from these, the porous lift-off approaches have received significant attention due to their wafer-scale porosification process followed by epitaxial growth or annealing. Porous-lift-off approaches generally consist of two routes, Germanium-on-Nothing (GeON or GON)77, and electrochemical porosification. These methods enable the formation of a uniform porous layer, ensure high-quality epitaxial growth, provide high solar cell performance, offer high throughput potential, and allow for multiple substrate reuse78. The GeON approach is based on lithography and plasma etching steps, and annealing to engineer a voided weak layer at the interface between the substrate and the Ge membrane/foil, allowing separation of Ge membrane/foil from the parent substrate. In this context, we define a foil as recrystalized Ge obtained through thermal annealing of Ge, whereas membrane refers to epitaxially grown crystalline Ge formed on top of a porous Ge layer. In contrast, electrochemical porosification route uses a simple and inexpensive method of bipolar electrochemical etching (BEE) to form porous layers. Two different routes have been explored, (1) Biporous (double porous) mesoporous Ge layer (pores 2–50 nm) followed by thermal annealing to form a recrystallized Ge foil45,79, and (2) the Porous Efficient Epitaxial LayEr Release (PEELER)15 based on single mesoporous layer and subsequent epitaxial growth of the Ge (Ge-epi) to form Ge nanomembrane (few hundred nanometer). Although in both processes, the Ge foil and the Ge membrane serve as growth templates for III-V layers. These techniques have attracted much attention recently owing to their rather simple, inexpensive and wafer-scale Ge membrane fabrication and multiple reuses of parent Ge wafer substrate15,78. Porosification routes: Process steps All three porosification-based approaches (GeON, Biporous mesoporous Ge layer, and single mesoporous Ge + Ge-epi) follow roughly five common process steps, as illustrated on the right-hand side of Fig. 1. First, a porous Ge layer (PGe) is formed: in the case of GeON, this is achieved by lithography and dry-etch patterning, whereas for Biporous mesoporous Ge layer and single mesoporous Ge layer + Ge-epi, bipolar electrochemical etching (BEE) is used. Next, the PGe layer undergoes thermal annealing. For GeON, this annealing step reorganizes the porous network to form a suspended monocrystalline Ge layer on top of a buried void, resulting in the so-called GeON structure. For Biporous mesoporous Ge layer, under thermal annealing, the bi-porous layer is reorganized into a porous Ge foil with a closed top surface and supported by pillars on the parent bulk substrate. In single mesoporous Ge layer + Ge-epi, before annealing, a thin (a few hundreds of nanometers) epitaxial Ge layer is grown. Subsequent annealing creates a mechanically weak interface between the bulk parent substrate and the epitaxial Ge, which enables the eventual detachment of a Ge membrane from the parent substrate. After preparing the engineered substrate, III–V active layers are epitaxially grown, followed by solar cell microfabrication80, including the mechanical detachment of the thin solar-cell structure81. Finally, the parent Ge wafer is reconditioned using a grinding + Chemical Mechanical Polishing (CMP) sequence for GeON82, only CMP for Biporous mesoporous Ge engineered substrate45,79, and a Wet-Etch-Reconditioning process demonstrated for single mesoporous Ge layer + Ge-epi, enabling multiple reuse cycles of the Ge substrate78. All of these steps are discussed in detail in the following sections. Porous Ge formation Porous lift-off approaches start with the porosification of Ge to form a PGe layer. Two different methods are reported in the literature for the formation of PGe layer. Porous Ge formation for Germanium-On-Nothing (GeON) The Silicon-on-Nothing (SON) architecture, first introduced by Jurczak et al.83, in 1999 demonstrated the feasibility of MOSFET integration and was later extended to CMOS (Jurczak et al., 2000)84 with ultra-thin buried dielectrics and silicon films via epitaxial processes. In this approach, buried cavities are formed beneath an ultra-thin silicon layer through selective removal of a sacrificial SiGe layer, creating a localized air gap that improves improved electrical performance compared to conventional silicon-on-insulator (SOI) substrates. Building on this concept, Batail et al. (2007)85 introduced GeON by combining SON substrates engineering with Ge condensation in SiGe layers, enabling ultrathin, high-Ge-content films on a buried dielectric with enhanced carrier mobility and electrostatic control. This process of forming monocrystalline Ge through condensation of SiGe on Si is further explored for thicker Ge layers, enabling applications in heterogeneous integration of various systems on SOI beyond MOSFETs86,87. In contrast, the Empty-Space-in-Silicon (ESS) technique for SON, first demonstrated by Mizushima et al. (2000)88 at Toshiba, forms buried cavities without sacrificial layer. The dependence of final SON structure on initial trench shape and layout was investigated by Sato et al. (2004)89 also at Toshiba, and the process was subsequently physically modeled by Sudoh et al. (2009)90. Instead of relying on the selective removal of a sacrificial SiGe layer, as in earlier approaches for SON formation, this process uses patterned silicon substrates with deep trenches, which undergo high-temperature hydrogen annealing, where surface diffusion driven by surface energy minimization (curvature effects) transforms the trench structures into spherical voids. Depending on the trench geometry (width, depth, and spacing between trenches), these voids can further evolve into pipe- and plate-shaped buried cavities. This self-organized process enables the formation of a continuous buried air gap beneath a crystalline silicon layer without requiring sacrificial material removal or wafer bonding. Building on the Empty-Space-in-Silicon method for SON, the concept of GeON approach introduces a top-down nanofabrication route for realizing buried void structure, first reported by Park et al.77. In this approach, the formation of cylindrical pores started by spin-coating a 200 nm thick nanoimprint resist on Ge substrate and baked at 130 °C for 5 min to remove residual solvent. Square arrays of cylindrical Si pillars served as the nanoimprint lithography (NIL) stamp, which was pressed into the resist under 40 bar pressures for 3 min using thermal NIL. After cooling to room temperature, the stamp was released, leaving a periodic array of openings in the resist. Deep reactive ion etching was then employed to create cylindrical pores in Ge, using alternating cycles of SF6 plasma etching and C4F8 passivation to precisely control the pore aspect ratio (AR). Finally, the resist and carbon residues were removed by O2 plasma treatment for 3 min, followed by rinsing in dilute ammonia solution for 10 min77. These cylindrical pores are annealed in hydrogen for reorganization into GeON structure. Final shape of GeON structure varied depending upon the initial morphology of Ge pores, which will be explained later in the section 4 (porous germanium transformation with annealing). It should be noted that, compared to electrochemical porosification (discussed in the following section), the GeON approach relies on standard semiconductor processing steps, including lithography, plasma etching, and cleaning, rather than electrochemical dissolution processes, and therefore offers better pore size controllability, albeit with a more equipment-intensive fabrication process. Porous Ge formation with Bipolar Electrochemical Etching (BEE) Literature reports on the formation of mesoporous Ge layer using unipolar (anodic) electrochemical etching unlike silicon (Si) yields an undesired oxidized surface instead of hydride terminated (Ge-Hx) germanium surface91,92. The first mesoporous Ge layer, of few microns thick with Ge-Hx terminated surface on both n-type and p-type Ge wafer, was reported by Buriak et al. (2000)93 using bipolar (anodic and cathodic) electrochemical etching (BEE) in an HCl-based electrolyte by continuously altering the anodic to cathodic bias; this current alternation was introduced so that the cathodic step passivates and protects the porous layer walls formed and allowed to have thicker layers without the dissolution of the already porous part. This mesoporous Ge layer exhibited notable resistance to oxidation but could undergo hydrogermylation reactions with alkenes and alkynes. Following this work, Giovanni et al. (2005)94 attempted the porosification of both n-type and p-type Ge in HF-based electrolytes. Although optimized anodization conditions allowed them to obtain a uniform porous film with a sponge-like structure on p-type Ge, the structures formed on n-type Ge were macroporous (pore size > 50 nm) and highly non-uniform, resulting in etch pits with depths and widths on the order of tens of microns. Subsequently, Fang et al. (2006)95 clarified that electrochemical pore formation in Ge involves a complex interplay between semiconductor physics and surface chemistry, fundamentally different from the behavior observed in Si. Their work consolidated experimental observations on both n-type and p-type Ge, highlighting challenges such as difficult pore nucleation, strong electropolishing tendencies, and unconventional crystallographic growth patterns. They further analyzed the influence of doping level, electrolyte composition, illumination, and surface conditioning on pore morphology, as well as phenomena such as oxide-filled pores and electrocapillarity effects. Building on these advancements, Rojas et al. (2009)96 achieved the reproducible formation of uniform and thick mesoporous Ge layer (with pores diameter ranging from 2 to 50 nm) on a 4-inch (edge excluded) p-type Ge wafer by electrochemical etching in highly concentrated HF electrolytes. Later, Rojas et al. (2011)97 demonstrated etching and lift-off of a porous Ge foil formed by porous layer formation via BEE, containing a porosity gradient, and subsequent hydrogen annealing to induce structural reorganization of the porous structure into a Ge foil supported by pillars. Following this work, Boucherif et al. (2013)98 investigated the double mesoporous Ge layer, comprising a low porosity (<30%) (low P) on top of high porosity (>70%) (high P) Ge porous layer instead of single mesoporous Ge formed by BEE. The bilayer structure was obtained by applying two different anodic currents, a low anodic current for low P and a higher current for high P, while maintaining the same cathodic current magnitude for both steps. Figure 2a shows, on the left the cross‑section scanning electron microscopy (SEM) image of the double mesoporous Ge layer, consisting of low P and high P regions, with respective porous layer thickness of 1.25 µm and 0.8 µm. On the right (i) and (ii) a high magnification SEM image, showing the morphological properties of low P and high P regions. The morphology was classified as “sponge-like” with the individual pore exhibiting a “curved cylindrical” shape and an average diameter of 7 nm98. Following this work, Tutashkonko et al. (2015)99 reported that by precisely tuning the surface passivation pulse duration during etching, mono-, bi-, and tri-layer structures with distinct morphologies can be obtained. Short passivation pulses (resulting in incomplete pore-wall passivation by =GeH2) lead to the formation of characteristic “pine-tree” or “fishbone” structures, where = GeH2 layers form preferentially at the bottom of the pore tips while the pore walls are covered with isolated GeH2 and =GeH–GeH= sites. Longer pulses, which passivate the pore walls completely, result in sponge-like layers. Later on, Schreiber et al. (2023)100 reported “fishbone‑like” morphology for low P layer and “sponge-like” morphology for high P layer and used this double mesoporous Ge layer as a template for III-V epitaxial growth.Fig. 2Morphology, uniformity and etching characteristics of porous Ge layers fabricated by bipolar electrochemical etching.SEM cross section Porous Germanium layer by using bipolar electrochemical etching (a) double porosity layer (low porosity (Low P) and high porosity (High P)), and high magnification SEM image of (i) Low P region and (ii) high P region (reproduced from98, with the permission of AIP Publishing, permission conveyed through Copyright Clearance Center, Inc.), (b) single porous layer (reproduced from15, with permission of Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.). Single porous layer wafer-scale uniformity for (c) porosity and (d) thickness (reproduced from15, with permission of Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.) and (e) etching rate and thickness variation as a function of current density (reproduced and adapted from103, licensed under CCBY 4.0). Morphology, uniformity and etching characteristics of porous Ge layers fabricated by bipolar electrochemical etching. SEM cross section Porous Germanium layer by using bipolar electrochemical etching (a) double porosity layer (low porosity (Low P) and high porosity (High P)), and high magnification SEM image of (i) Low P region and (ii) high P region (reproduced from98, with the permission of AIP Publishing, permission conveyed through Copyright Clearance Center, Inc.), (b) single porous layer (reproduced from15, with permission of Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.). Single porous layer wafer-scale uniformity for (c) porosity and (d) thickness (reproduced from15, with permission of Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.) and (e) etching rate and thickness variation as a function of current density (reproduced and adapted from103, licensed under CCBY 4.0). Also, Tutashkonko et al. (2013)101 not only reported thick mesoporous Ge layer with diverse morphologies (sponge like, columnar and fishbone) formed by BEE but also showed that the duration of anodic and cathodic pulses, the applied current density and the total anodization time strongly influence the mesoporous Ge morphology, producing film thicknesses from approximately 300 nm up to around 7 µm over three hours. To achieve comparable thicknesses in far less time, Bioud et al. (2017)102 introduced fast bipolar electrochemical etching, which combines high current densities with short pulse durations BEE (also known as high frequency BEE). This method delivers a growth rate up to 300 nm per minute, forming 6 µm‑thick homogeneous mesoporous Ge layers in just 40 min instead of three hours. Additionally, an optimized nucleation step is introduced to the process. This step consists of applying a direct anodic current for 30 s before alternating the bias in order to ensure a uniform distribution of the current over the surface of the wafer so that lateral uniformity of the porous layer is improved. Following the work of Bioud et al. (2017), Hanus et al. (2023)103 demonstrated the wafer-scale formation of homogeneous mesoporous Ge layers by optimizing the BEE process. A custom-designed porosification cell (discussed later in this section) enabled edge-to-edge porosification, increasing wafer’s usable area. Following their wafer-scale demonstration of mesoporous Ge layer formation, Paupy et al. (2023)15 presented the PEELER (single mesoporous Ge layer + Ge-epi) process to fabricate detachable, wafer-scale monocrystalline Ge nanomembrane (700 nm thick) and also demonstrated successful substrate reconditioning (surface quality suitable for III-V solar cells), enabling its reuse over multiple cycles. The one-step single-layer process (Fig. 2b) appears simpler than the earlier double-porous layer methods of Schreiber et al.100 and Boucherif et al.98, in terms of cutting down both processing time and resources used15. However, the single-porous approach requires a subsequent Ge epitaxial growth step, which is more complex and costly than the simple annealing run used in the bi-porous approach. The resulting single mesoporous Ge layer nevertheless showed excellent uniformity in both thickness (Fig. 2c) and porosity (Fig. 2(d)) with variation below 2%15. Recently, Zhu et al. (2023)104 using bipolar potentiostatic etching obtained porous Ge layer with the thickness of more than 100 nm and pore diameter in the range of 10 to 100 nm for both p- and n-type Ge-on-Si without illumination. As illumination is often required to generate additional holes, particularly for facilitating n-type porosification, which will be explained in more details in the following section. The mechanism of bipolar electrochemical etching (BEE) Unlike electrochemical etching on silicon (Si) substrates, the application of a direct anodic current to Ge for a period of time allows charges and ions (F­; H+; Ge) transfer between an electrolyte (aqueous hydrofluoric acid (HF)) and Ge, leading to the formation of germanium dioxide (GeO2)99. This GeO2 easily dissolves in aqueous solution and hence electrochemical etching of Ge occurs. To prevent this Ge dissolution and to form a porous Ge layer, a subsequent cathodization step after the initial anodization is performed. This alternating anodic-cathodic protocol is referred as bipolar electrochemical etching (BEE)93. In BEE, electrochemical reactions contain two parts: anodic dissolution and cathodic passivation. Depending on the types of applied potential, two types of anodic electrochemical dissolution of Ge in HF electrolytes exist102,105: The divalent dissolution, which takes place below 0.15 V:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Ge}}}}+6{{{\rm{HF}}}}+{{{{\rm{h}}}}}^{+}\to {{{\rm{Ge}}}}{{{{\rm{F}}}}}_{6}^{2-}\,+4{{{{\rm{H}}}}}^{+}+{{{{\rm{H}}}}}_{2}+{{{{\rm{e}}}}}^{-}$$\end{document}Ge+6HF+h+→GeF62−+4H++H2+e− The tetravalent dissolution, which takes place above 0.15 V:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Ge}}}}+6{{{\rm{HF}}}}+{{{{\rm{h}}}}}^{+}\to {{{\rm{Ge}}}}{{{{\rm{F}}}}}_{6}^{2-}\,+6{{{{\rm{H}}}}}^{+}$$\end{document}Ge+6HF+h+→GeF62−+6H+Here, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${h}^{+}$$\end{document}h+ is the hole in the valence band and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${e}^{-}$$\end{document}e− is the electron in the conduction band. As it can be seen from the equations that holes play a key role in both dissolution mechanisms, essential for the primary steps and limiting the dissolution rate. During the anodic step, holes generated in the Ge bulk migrate to the surface, rendering the chemisorbed Ge–H sites electrophilic. Fluoride ions (F⁻) then attack these sites in a nucleophilic substitution, forming Ge–F bonds. The high electronegativity of fluorine withdraws electron density from Ge, weakening its lattice attachment in the lattice and enabling successive F⁻ attacks that release Ge atoms as GeF4102:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\rm{Ge}}}}+4{{{\rm{F}}}}+{4{{{\rm{h}}}}}^{+}\to {{{\rm{Ge}}}}{{{{\rm{F}}}}}_{4}$$\end{document}Ge+4F+4h+→GeF4 This GeF4 quickly transforms in HF electrolyte to the H2GeF6 compound following:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{{\rm{GeF}}}}}_{4}+2{{{\rm{HF}}}}\to {{{{\rm{H}}}}}_{2}{{{\rm{Ge}}}}{{{{\rm{F}}}}}_{6}$$\end{document}GeF4+2HF→H2GeF6 Hence, the surface of Ge under anodic bias ends up covered with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\equiv {{{\rm{Ge}}}}-{{{\rm{OH}}}}$$\end{document}≡Ge−OH groups unlike a chemically stable hydrogenated surface. As a result, current flows not only through the pore tips but also along the pore sidewalls, leading to thinning and ultimately collapsing of the original porous structure as shown in left hands side of Fig. 3 (a)102.Fig. 3Electrochemical porosification mechanisms and porosification cell configuration for porous Ge fabrication.a Electrochemical porosification mechanism of Ge in HF medium: (left) direct current: (1) H-passivated Ge in HF without anodic current, (2) F- ions attack Ge-H bonds, (3) F-passivated Ge in the presence of holes, (4) Pulled out Ge from the bulk, (5) Ge-Ge bonds breaking and GeF4 molecule release, (6) GeF4 transformation in HF electrolytes, (7) OH-passivated Ge in the presence of holes; (right) alternating current (BEE technique): (1) H-passivated Ge, (2–7 steps) anodic etching, (8) repulsed negative charges, (9) cathodic passivation in the presence of electrons and Ge-H bond formation (reprinted from102, with permission from Elsevier). b Diagram of porosification cell (reproduced from US Patent Application US20240376629A1 (2024))111. c Illustration of the porosification cell (reproduced from103, licensed under CCBY 4.0). Electrochemical porosification mechanisms and porosification cell configuration for porous Ge fabrication. a Electrochemical porosification mechanism of Ge in HF medium: (left) direct current: (1) H-passivated Ge in HF without anodic current, (2) F- ions attack Ge-H bonds, (3) F-passivated Ge in the presence of holes, (4) Pulled out Ge from the bulk, (5) Ge-Ge bonds breaking and GeF4 molecule release, (6) GeF4 transformation in HF electrolytes, (7) OH-passivated Ge in the presence of holes; (right) alternating current (BEE technique): (1) H-passivated Ge, (2–7 steps) anodic etching, (8) repulsed negative charges, (9) cathodic passivation in the presence of electrons and Ge-H bond formation (reprinted from102, with permission from Elsevier). b Diagram of porosification cell (reproduced from US Patent Application US20240376629A1 (2024))111. c Illustration of the porosification cell (reproduced from103, licensed under CCBY 4.0). Switching to cathodic step, a high negative current repulse all the negative charges at the electrolytes/Ge interface (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{{\rm{F}}}}}^{-}{,{{{\rm{OH}}}}}^{-}$$\end{document}F−,OH−) and promotes the hydrogenation of the surface as follows102:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\equiv {{{{\rm{Ge}}}}({{{\rm{OH}}}})}_{2}+4{{{{\rm{e}}}}}^{-}+4{{{{\rm{H}}}}}^{+}\to {{{{\rm{GeH}}}}}_{2}+{{{{\rm{H}}}}}_{2}{{{\rm{O}}}}$$\end{document}≡Ge(OH)2+4e−+4H+→GeH2+H2O The hydrogen termination of Ge presents an energy barrier across the Ge electrolyte interface for the charge transfer (right hand side of Fig. 3a) and hence passivates the porous germanium layers formed during anodization step, which otherwise would have dissolved. This anodic–cathodic cycling enables a fully reversible transformation of the Ge surface from a hydroxylated ( ≡ Ge–OH) state to a hydrogenated ( ≡ Ge–H) one. Under cathodic bias, the surface becomes hydrogen-terminated and strongly hydrophobic, whereas under anodic bias, it becomes hydroxyl-terminated and hydrophilic. This voltage-induced change in surface properties at Ge-electrolyte interface is referred to as the “electrocapillarity” of Ge106. In fast bipolar electrochemical etching, the anodic step duration kept too short compared to cathodic step duration to passivate the Ge porous layer and prevent pore walls dissolution while to keep the pore growth at the pore tip102. Influence of process parameters on porous layer formation Several parameters, including applied current density, anodic/cathodic frequency, and pulse duration, type of Ge wafer (whether n-type or p-type), play a critical role in determining the characteristics of the porous layer, such as its etching rate, porosity, uniformity and morphology. For example, it is shown for p-Ge the etching rate varies linearly with applied current density, allowing higher etching rates with increasing the current density, (y-axis on right-hand side of Fig. 2e)103. At the same time, the porosity of the p-Ge layer was found to be fine-tuned with increasing applied current density, indicating porosity modulation with applied current density, (y-axis on left-hand side of Fig. 2e)103. As stated previously, high switching frequency between anodic and cathodic bias in fast bipolar electrochemical etching regime gives uniform porous Ge layer formation due to effective protection of pores walls and stimulus etching at the pore tips102. Also, it was shown that the duration of the anodization should be equal to or shorter than the cathodization duration, as this is a prerequisite for passivation of porous layer formed during cathodization step101. The application of initiation current (direct current before applying the alternating current) improved the current distribution across the wafer surface, leading to more uniform pore distribution and growth102. Porosification in relation to the doping of Ge has been only sparsely studied. Although the first report of BEE included both n-type and p-type Ge93, subsequent studies on n-type Ge porosification have been limited. One exception, published a few years after the initial report, investigated the nucleation and growth of porous Ge in both n-type and p-type substrates while considering various electrolytes, crystal orientations, and doping concentrations95. More recent work has also explored porosification in both n-type and p-type Ge-on-silicon substrate104. Unlike earlier approaches that required illumination to generate holes in n-type Ge, this work has shown that bipolar potentiostatic etching enables the formation of porous Ge in both p-type and n-type Ge-on-Si without the need for illumination. This method relies on periodically switching between anodic and cathodic voltages. In n-type Ge, applying a sufficiently high anodic voltage induces surface inversion, generating the required holes at the interface for electrochemical etching. Additionally, threading dislocations in n-type Ge epitaxial layer introduce intermediate electronic states that accumulate holes to participate in anodic etching reaction, enabling porosification even in the absence of external illumination. Using this approach, uniform porous layers exceeding 100 nm with pore sizes of 10–100 nm were successfully produced in low-doped n-type Ge-on-Si. It should be noted, however, that the porosification behavior of n-type Ge differs fundamentally from that of p-type Ge due to limited hole availability, which are required for anodic dissolution. In n-type Ge, electrons are the majority carriers, while holes are scarce under normal conditions. Consequently, additional excitation, such as strong electric fields and/or illumination is typically needed to generate electron–hole pairs and enable effective electrochemical etching. Consequently, pore growth in n-type Ge is generally more difficult to control, often leading to irregular morphologies, branching, and slower dissolution rates. These challenges explain why most early progress in porous Ge was achieved using p-type substrates and highlight the significance of recent bipolar potentiostatic etching on n-Ge-on-silicon104 as an effective strategy for enabling controlled porosification of n-type Ge. In contrast, for p-type Ge, holes are the majority carriers and are abundantly available at the surface, making uniform porous formation relatively easy even without backside illumination107; however, the lack of illumination also leads to slower etching kinetics, resulting in a comparatively slow formation of porous Ge102. It is important to mention here that for both p- and n-type Ge, the material pore density, pore dimension and layer structure depend on the doping density and the crystallographic orientation of the wafer102,104. Regarding etching directions, the crystallographic orientation of the wafer significantly influences the etching, as some direction poses more resistance towards etching than the others. For instance, (100) planes are etched faster than the (111) planes95,102. Lastly, the hydrogen gas evolution that is formed during electrochemical etching of Ge can affect the structure of porous layer integrity and its release enables homogeneity of porous layer, which otherwise damage the walls of layer and leading to the electropolishing of Ge103. Porosification equipment Multiple sorts of porosification equipment have been developed for porosification of silicon, such as those produced by AMMT108 and THERMCOSYSTEMS109. Although some of these silicon porosification tools are compatible for bipolar process required for Ge, they nevertheless prevent edge porosification through effective edge masking. Consequently, most groups have designed custom porosification cells for Ge porosification without edge masking. The only known cell that is capable of porosification of Ge edges included is developed by Kymosis110, a spin-off from the Université de Sherbrooke (UdeS) based on the work of Provost et al., as shown in the Fig. 3b111. The company had been actively advancing solutions for Ge porosification. Their innovative design enables efficient and uniform porosification while minimizing complexity and contamination risks, making it particularly suited for bipolar etching processes used for Ge. The working principle of the porosification cell is that a generator (or a potentiostat) runs a current through a semiconductor-electrolyte junction by connecting the back side of a Ge wafer “working electrode” and an ideally featureless counter electrode in the electrolyte as shown in Fig. 3b. The flow of current allows the transfer of charge and ions between the electrolyte and Ge wafer, leading to the formation of mesoporous Ge as explained in the “The mechanism of bipolar electrochemical etching (BEE)” section. It is important to mention here that, the choice of material for working electrode is crucial for ensuring effective and uniform etching. There are different types of materials used as electrodes for porosification of Ge substrate111Working electrode: The typical electrode material used for working electrodes are Copper (Cu) and glassy carbon.Counter electrode: Platinum (Pt) due to its high chemical inertness and electrochemical stability, it is widely used as a counter electrode material. Its resistance to corrosion and ability to operate in various electrochemical environments make it a preferred choice. Working electrode: The typical electrode material used for working electrodes are Copper (Cu) and glassy carbon. Working electrode: The typical electrode material used for working electrodes are Copper (Cu) and glassy carbon. Counter electrode: Platinum (Pt) due to its high chemical inertness and electrochemical stability, it is widely used as a counter electrode material. Its resistance to corrosion and ability to operate in various electrochemical environments make it a preferred choice. Counter electrode: Platinum (Pt) due to its high chemical inertness and electrochemical stability, it is widely used as a counter electrode material. Its resistance to corrosion and ability to operate in various electrochemical environments make it a preferred choice. Nonetheless, the selection of material for working electrode depends on the specific porosification process conditions and the desired properties of the porous layer. Finally, the materials and design of the cell are specifically selected to ensure high resistance to aggressive chemicals such as HF, enhancing both operator safety and equipment durability in industrial environments. Porous germanium transformation with annealing The thermal transformation of porous Ge (PGe) plays a pivotal role in the fabrication of Ge membrane as it induces the structural transformation necessary for mechanical stability and detachment. The annealing process induces atomic rearrangement, allowing Ge atoms to migrate into energetically favorable positions, thereby minimizing the surface energy112. This controlled morphological evolution tunes the PGe metrics; both porosity and thickness, which is essential for tailoring the mechanical and electrical properties112. Furthermore, thermal treatment promotes the formation of a weak void layer, a crucial feature that enables freestanding membrane production113. However, perhaps even more of importance, thermal treatment should allow annealing to tune the adhesion strength of the Ge membrane strong enough to support III–V growth yet weak enough for easy detachment after processing45. Thermally induced transformation of single-porosity mesoporous Ge layer Temperature-dependent morphological evolution Morphological changes of sponge-like mesoporous Ge begins at temperature (T) as low as 300 °C, marked by the formation of elongated cavities through surface diffusion (the movement of atoms along the pore walls) and their merging ultimately led to the development of spherical cavities, which have lowest surface energy, Fig. 4a–c114. Since the surface energy in Ge is anisotropic (dependent on crystallographic orientation), it can be further minimized through faceting. At 450 °C, cavity facets bounded by (100) and (111) crystallographic planes, which possess the lowest surface energy, emerge, further stabilizing the structure114. A similar physical phenomenon during thermal annealing of a low porosity (<40%) Ge layers was reported, where small voids migrate towards larger voids in underlying layer, driven by minimization of specific surface energy98. Even with temperature increase to 600–700 °C, the voids undergo faceting, aligning with specific crystallographic planes while preserving the crystalline order of the mesoporous Ge layer115. However, for the high porosity (>40%) surface energy minimization does not follow crystallographic orientation116, but instead follows an increase in crystallite size (Fig. 4d) where small crystallites agglomerates into large ones, as shown in Fig. 4e–h116. Such a transformation of crystallite size with annealing temperature obeys the Arrhenius-like relation112:6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\mathrm{ln}}}}{({dcrystallite})}=-\frac{{E}_{a}}{k}\cdot \frac{1}{T}+{{{\mathrm{ln}}}}(A)$$\end{document}ln(dcrystallite)=−Eak⋅1T+ln(A)Fig. 4Microstructural evolution of porous Ge during thermal annealing. Different types of the cavity shapes observed during the annealing.At the beginning of the thermal treatment, (a) elongated cavities formed by the coalescence of two or more neighboring pores and (b) spherical cavities are observed. At later stages, (c) faceted cavities start to dominate. The measured angle between the two most prominent orientations is about 55° (reproduced from114, with the permission of AIP Publishing, permission conveyed through Copyright Clearance Center, Inc.) d Crystallite size variation as a function of annealing temperature (10 min annealing), (e–h) cross-sectional SEM images showing morphological evolution of mesoporous Ge annealed for 10 min at increasing temperatures: as-prepared, 400 °C, 500 °C, and 600 °C, (i) time-dependent annealing at different temperatures, and cross-sectional SEM images showing morphological evolution of mesoporous Ge annealed at 500 °C for varying durations: (j–m) 0 min, 1 min, 30 min, and 300 min (adapted and reprinted116, with permission from Elsevier). Microstructural evolution of porous Ge during thermal annealing. Different types of the cavity shapes observed during the annealing. At the beginning of the thermal treatment, (a) elongated cavities formed by the coalescence of two or more neighboring pores and (b) spherical cavities are observed. At later stages, (c) faceted cavities start to dominate. The measured angle between the two most prominent orientations is about 55° (reproduced from114, with the permission of AIP Publishing, permission conveyed through Copyright Clearance Center, Inc.) d Crystallite size variation as a function of annealing temperature (10 min annealing), (e–h) cross-sectional SEM images showing morphological evolution of mesoporous Ge annealed for 10 min at increasing temperatures: as-prepared, 400 °C, 500 °C, and 600 °C, (i) time-dependent annealing at different temperatures, and cross-sectional SEM images showing morphological evolution of mesoporous Ge annealed at 500 °C for varying durations: (j–m) 0 min, 1 min, 30 min, and 300 min (adapted and reprinted116, with permission from Elsevier). where, Ea is the activation energy that equals to 0.69 eV for this mass-diffusion process, k is the Boltzmann constant, A is a proportionality constant and T is absolute temperature in Kelvin. For the mesoporous Ge layer with even higher porosity (72%), in a same tendency, thermal annealing leads to structural evolution, governed by Ostwald ripening as predicted by Lifshitz-Slyozov Wagner (LSW) theory, which will be explained later in this section. This evolution impacts both crystallinity and electrical conductivity. With annealing (T < 450 °C) no significant changes observed in the morphology while at T > 450 °C strong morphological changes observed from porous material to granular structure combined with a loss in long range crystalline order112. Time-dependent structural evolution Similar to temperature-dependent morphological evolution, annealing of the mesoporous Ge samples at a constant temperature for different durations (depending on the annealing temperature (Fig. 4i) leads to a morphology evolution. However, it is reported that for a given thickness of mesoporous Ge layer, lower annealing temperatures require longer annealing times to completely transform the porous structure into a stable void structure, while preserving the monocrystalline nature of the material. Additionally, the time required for the mesoporous Ge layer to shrink (i.e., reduction in layer thickness and transformation into monocrystalline Ge) depends nonlinearly on its initial thickness114. The impact of time on structural reorganization is less significant than that of temperature. Major reorganization of crystallite size occurs during the first few minutes of the annealing (Fig. 4j–m), followed by a slower rate of change at all T [400–600 °C] but it’s faster in the case of high T, following a simple power law116:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${d}^{2}-{d}_{0\,}^{2}=K\times t$$\end{document}d2−d02=K×t Where: d is the effective crystallite size of porous Ge network (i.e., the characteristics size of the solid Ge region) rather than the size of individual pore. The parameter d0 is the initial crystallite size before annealing, t is the annealing time, K is a positive and time-independent constant associated with thermally driven growth of crystallite size during annealing. The increase in crystallite size after annealing confirms that K is positive. Thermally induced morphological evolution of double-porosity mesoporous Ge The double-porosity mesoporous Ge layer or bilayer is formed of a low-porosity layer serving as an epitaxial growth template on top of high-porosity one used as a weak separation layer. Thermal annealing with increasing temperature affects both porous layers simultaneously; the high-porosity layer transforms into large crystallites serving as pillars connecting the substrate to the rest of structure. While low-porosity layer acts as a template for epitaxial growth with Swiss-cheese morphology98,117, as shown in Fig. 5a–c98. It is worth mentioning that the top surface of low-porosity layer becomes fully closed with annealing, thereby providing a continuous template for subsequent III-V epitaxial growth. The total thickness stability (thickness of high-porosity and low-porosity layer) can be achieved through a slight shrinkage in high-porosity layer in the favor of low-porosity layer. Meanwhile, varying the annealing time reveals the step-by-step formation of pillars and the densification of the low-porosity layer. Finally, at the end of reorganization process, pillars connecting the thin film to the bulk are formed that helps in the subsequent lift-off process98.Fig. 5Temperature-dependent morphological evolution of mesoporous double-layer Ge during annealing.Cross-sectional SEM images showing the resulting morphology of the mesoporous double-layer structure for 10 min annealing in vacuum at annealing temperatures: (a) 600 °C; (b) 650 °C and (c) 700 °C (reproduced from98, with permission of AIP publishing, permission conveyed through Copyright Clearance Center, Inc.) Temperature-dependent morphological evolution of mesoporous double-layer Ge during annealing. Cross-sectional SEM images showing the resulting morphology of the mesoporous double-layer structure for 10 min annealing in vacuum at annealing temperatures: (a) 600 °C; (b) 650 °C and (c) 700 °C (reproduced from98, with permission of AIP publishing, permission conveyed through Copyright Clearance Center, Inc.) Thermally induced morphological evolution of GeON Similarly, thermal annealing of cylindrical pores arranged in square arrays on Ge substrate (fabricated via nanoimprint lithography and reactive ion etching) leads to the reorganization of Ge into a GeON structure. The final shape of GeON depends upon the pore morphology such as pore diameter, pitch (distance between adjacent pores), and depth-to-diameter aspect ratio (AR). Each pore transforms into a spherical void when the initial aspect ratio (AR) is sufficiently high. Furthermore, when the diameter of these voids exceeds the initial pore pitch, neighboring voids coalesce to form a continuous plate-shaped void beneath a suspended Ge layer. Figure 6a, c shows the SEM images of initial pores, where pitch (200 nm) and diameter (145 nm), correspond to a porosity of 41.3%, (calculated by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left(\frac{\pi {D}^{2}}{4{P}^{2}}\right)$$\end{document}πD24P2 where, D and P are the diameter and pitch of pores, respectively), are kept constant, while AR is varied (4.2, 5.2 and 7.2). Figure 6d–f show the corresponding GeON morphologies after annealing at 730 °C for 60 min in hydrogen under ambient pressure, at which reorganization becomes active.Fig. 6Pore reorganization in macroporous Ge structure.a–c SEM images of initial Ge pores with aspect ratios of 4.2 (a), 5.2 (b), and 7.2 (c) and (d–f) corresponding SEM images of reorganized pores of (a–c), respectively (reprinted from77, with permission from Elsevier). Pore reorganization in macroporous Ge structure. a–c SEM images of initial Ge pores with aspect ratios of 4.2 (a), 5.2 (b), and 7.2 (c) and (d–f) corresponding SEM images of reorganized pores of (a–c), respectively (reprinted from77, with permission from Elsevier). As it can be seen Ge pores with AR of 4.2 transformed into spherical void after annealing with a diameter of 175 nm (Fig. 6a, d). As noted earlier, this morphology evolution is analogous to Rayleigh-like instability (discussed in the following section) and is primarily driven through the surface diffusion of Ge along the pore walls to minimize the surface energy. As the pore aspect ratio (AR) increases during annealing, the void diameter grows and approaches the initial pore pitch. Consequently, neighboring spherical voids begin to coalesce. When the void diameter becomes comparable to the pore pitch, partial merging occurs; for example, pores with an AR of 5.2 exhibit partial coalescence after reorganization (Fig. 6b, e). The reformed, merge voids becomes horizontally elongated ellipsoids due to stability of the (100) facet, Fig. 6b, e77. If the AR of Ge pores further increase to 7.2, the reformed voids merges completely with each other and form a thin Ge layer with thickness of 320 nm on a plate-shaped void, Fig. 6c, f. Additionally, if the initial porosity is too high, the pore inlets fail to close due to insufficient mass transport of Ge from the pore walls during annealing, leading to collapse of the cylindrical pores and resulting in high surface roughness. In contrast, when the porosity or aspect ratio (AR) is too low, the structure evolves into isolated spherical GeON voids or only partially coalesced morphologies. Therefore, to achieve a continuous plate-shaped void beneath a GeON layer—suitable for device epitaxy and layer transfer—an optimal range of initial pore parameters is required, typically porosities of ~35–45% and AR values of ~5.5–10. Furthermore, the thickness of the resulting GeON layer can be tuned by adjusting the initial pore morphology, particularly the aspect ratio (AR) and pitch, while an optional epitaxial step may be employed to achieve the desired thickness. It is demonstrated also that the relationship between GeON layer thickness and the initial AR was consistent with the empirical model proposed by Sato et al.89, for predicting the morphology of ESS structures in SON. To sum-up, both the final GeON morphology and film thickness can be systematically controlled through careful design of the initial pore geometry. Finally, important to mention here that the atmosphere has a noticeable effect on the reorganisation of PGe. Annealing in hydrogen, nitrogen or argon atmospheres or under vacuum, yields different structural reorganization of the PGe layer. Annealing in hydrogen atmosphere gives a stable PGe layer structure for further processing77, while annealing under vacuum can induce faster reorganization compared to other atmospheres, enabling the shrinkage of the PGe layer114. Underlying phenomenon (Ostwald ripening and LSW theory) Understanding the morphological evolution of PGe layers during annealing is essential for controlling membrane formation. This nano- and microscale evolution of PGe layers follows the Ostwald ripening phenomenon, where small cavities/pores shrink and dissolve, while larger ones grow, leading to surface energy minimization. This process is quantitatively described by LSW theory, treating as-prepared PGe as a nano-hollow crystal supersaturated with vacancies, and describes evolution of these vacancies to attain a final equilibrium state. The shrinkage/dissolution or growth of cavity is defined by the cavity critical radius (RC). The RC is the minimum radius at which vacancy flow stops due to the same vacancy concentration inside the cavity and in the vacancy field (entire set of cavities defines an average level of vacancy supersaturation in the lattice)114. Hence, cavity evolution is governed by the critical radius as follows: a.If R > RC the cavities grow b.If R < RC the cavities shrink/dissolve The cavities close to the surface can transfer their vacancies to large neighbor cavities, making the surface to act as a vacancy sink, forming a non-porous close surface. The evolution of RC with annealing temperature T and time duration t follows the equation114:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{c}(t) \sim \left({\frac{4{Dt}}{9T}}\right)^{\frac{1}{3}}$$\end{document}Rc(t)~4Dt9T13Where, D is vacancy diffusion coefficient (follows Arrhenius dependence), T is annealing temperature (K), and t is annealing time. This Ostwald-driven transformation controls pore coalescence, void formation, and the closure of the top surface, all of which are critical for producing mechanically stable Ge membrane/foil that serves as suitable templates for III-V epitaxial growth. By understanding this mechanism, annealing conditions can be optimized to achieve the desired nanostructure and mechanical integrity. It is important, however, to note that describing the evolution of porous Ge network in terms of crystallites growth or cavity/pore depends on the porosity of PGe. At lower porosity (typically < ~45%), the structure is dominated by interconnected pores, and pore evolution is better described by Eq. (8). In contrast, at higher porosity ( > ~55%), the structure consists of isolated or weakly connected crystallites, and the evolution of crystallite size (Eq. (7)) becomes more relevant116. The transition between these regimes depends on porosity, however, the exact threshold values may vary depending on techniques used to form the porous Ge layer. The approximate porosity ranges indicated here are based on empirical observations and are consistent with general trends reported in the literature112,114 on thermally induced reorganization of porous Ge. Therefore, these two descriptions of treating PGe are not contradictory but rather complementary, detailing different aspects of the thermally induced reorganization process. III-V epitaxial growth on PGe The III-V epitaxial growth on PGe can be broadly categorized into two types, depending on the nature of engineered Ge substrate used. III-V epitaxy on Ge foil derived from macroporous Ge substrate The macropores are characterized by pore diameters greater than 50 nm, featuring larger and organized void structures. In this method, GeON approach is used to engineer the Ge substrate upon which III-V epilayers are grown. The GeON concept, based on a lithography and dry-etching patterning process, was initially developed as a proof of concept for structuring Ge surfaces77. The core mechanism of GeON involves the controlled restructuring of macroporous Ge at high temperatures, forming a GeON structure (suspended Ge foil attached by a few anchor points to the wafer) with an estimated practical thickness of 5 µm118, This thickness is on lower side for efficient infrared light absorption for solar cell applications. However, the GeON template can be thickened through chemical vapor deposition (CVD) using precursors such as GeCl4 or GeH4, particularly if the goal is to create a solar cell with an active Ge junction119. As shown in Fig. 7a-ii, a ~ 15 µm Ge epilayer was grown on a GeON foil to enable wafer-scale upscaling. SEM and X-ray diffraction (XRD) analyses were conducted to confirm that thickening did not degrade the morphology or crystal structure. Triple-junction solar cells were subsequently grown on the thickened Ge using MOCVD, with the Ge serving as the bottom cell. The reported performance of small-area solar cells was comparable to that of cells grown on standard Ge wafers119.Fig. 7Epitaxial growth of Ge and III-V layers on GeON templates and mesoporous Ge structure.a GeON seed after dry-etching: (i) top view photograph of after III-V epitaxy on GeON and (ii) SEM images of GeON structure (reproduced and adopted from119, with the permission of © 2023 John Wiley & Sons Ltd, permission conveyed through Copyright Clearance Center, Inc.). b SEM image of III-V epitaxy on biporous Ge layer (reprinted from117 with permission from Elsevier). c Single mesoporous Ge + Ge-epi: (i) optical image of an epitaxial Ge grown on mesoporous Ge/Ge at the wafer-scale, (ii) TEM image of mesoporous Ge + Ge-epi structure (reproduced from15, with the permission from Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.). (iii) schematic illustration of Ge-epi growth on mesoporous Ge substrate, 3D island growth, coalescence of islands into Ge membrane and thickening of the Ge membrane by 2D layer-by-layer growth (reproduced from46, licensed under the CCBY 4.0) (iv-v) top view photograph and cross SEM after III-V epitaxy on single mesoporous Ge + Ge-epi (reproduced from80, licensed under the CCBY 4.0). Epitaxial growth of Ge and III-V layers on GeON templates and mesoporous Ge structure. a GeON seed after dry-etching: (i) top view photograph of after III-V epitaxy on GeON and (ii) SEM images of GeON structure (reproduced and adopted from119, with the permission of © 2023 John Wiley & Sons Ltd, permission conveyed through Copyright Clearance Center, Inc.). b SEM image of III-V epitaxy on biporous Ge layer (reprinted from117 with permission from Elsevier). c Single mesoporous Ge + Ge-epi: (i) optical image of an epitaxial Ge grown on mesoporous Ge/Ge at the wafer-scale, (ii) TEM image of mesoporous Ge + Ge-epi structure (reproduced from15, with the permission from Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.). (iii) schematic illustration of Ge-epi growth on mesoporous Ge substrate, 3D island growth, coalescence of islands into Ge membrane and thickening of the Ge membrane by 2D layer-by-layer growth (reproduced from46, licensed under the CCBY 4.0) (iv-v) top view photograph and cross SEM after III-V epitaxy on single mesoporous Ge + Ge-epi (reproduced from80, licensed under the CCBY 4.0). III-V epitaxy on Ge foil derived from mesoporous Ge substrate III-V epitaxy on single mesoporous Ge The first attempt to grow III-V (GaInAs) on mesoporous Ge (via BEE) using metal-organic vapor phase epitaxy yielded solar cell device efficiency of 7.7%, compared to 16.2% on bulk Ge, due to high shunt defects from the rough, coral-like surface of the porous structure120. Protruding defects (coral-like structure at the surface) formed during annealing of the mesoporous Ge layer (750 °C, 60 min) were addressed using a post-annealing double-surface treatment (HBr dip followed by ultrasonic DI water), and thereby significantly improved surface quality. This surface treatment enabled monocrystalline GaInAs growth, confirmed by wide-angle XRD and Transmission Electron Microscopy (TEM). Notably, the embedded porosity was preserved, allowing for possible layer lift-off. This study served as a first proof-of-concept that mesoporous Ge can be restructured for functional III-V device integration. However, further work was needed to eliminate surface protrusions, achieve clean release and regrowth, and improve device performance. III-V epitaxy on Biporous mesoporous Ge substrate The first reported attempt to grow III–V epitaxy on an annealed biporous mesoporous Ge substrate (800 °C for 30 min) employed an Al0.5Ga0.49In0.01As/Ga0.99In0.01As double heterostructure, grown via metal-organic vapor phase epitaxy, as shown in Fig. 7b117. XRD analysis confirmed that the composition and thickness of the III–V layers were comparable to those grown on a conventional Ge reference substrate. Atomic Force Microscopy (AFM) revealed similar surface morphology before and after the epitaxy for both the substrate. Moreover, time-resolved photoluminescence measurements indicated a minority carrier lifetime of 7.6 ns in the low injection regime, demonstrating that the electrical quality of the III–V layers on biporous Ge was comparable to that on standard Ge substrates. In another attempt, a sintering process at 720 °C for 5 min was used to transform the structure, where low-porosity layer densified into a porous Ge foil and the high-porosity layer remained as a weak interface, enabling efficient layer lift-off100. The crystalline quality of the annealed double-porous mesoporous Ge layer was confirmed by reciprocal space mapping, which showed no change in the in-plane lattice parameter of the substrate as a result of either porosification or annealing. Thus, these efforts indicates the potential of the biporous Ge layer as a template for the high crystalline quality growth of III-V based MJSCs, layer lift off and reuse of the wafer45. While earlier publications did not report detailed device performance, recent preliminary solar cell results presented at a recent conference (IEEE PVSC 54) show that III–V layers on bi-porous Ge substrates have comparable quality and external quantum efficiency to bulk Ge, with only a slightly lower open-circuit voltage (VOC) of 60 mV, likely due to surface defects79. III-V Epitaxy on single mesoporous (Ge + Ge-epi) substrate This approach involves two-step epitaxial growth on single mesoporous Ge substrate: first, a Ge epilayer is grown on mesoporous Ge, followed by the growth of a III-V layer on top of Ge-epilayer15,45. Ge-epi on mesoporous PGe The Ge epitaxial growth on mesoporous Ge was done using a molecular beam epitaxy (MBE) reactor under a high vacuum (10−6 torr)15,46. A solid source of Ge was used with a K-cell temperature at 1250 °C and the growth was carried out in two steps. First, a 200 nm thick Ge buffer layer is deposited at low temperature of 200 °C to protect porous structure and ensure crystalline seeding. Then, the 500 nm thick high temperature 475 °C Ge layer deposited, providing the template for III-V epilayer for device integration. After the growth, the visual inspection of the 100 mm Ge nanomembrane shows a mirror-like surface (Fig. 7c-i15, and the surface quality of Ge epilayer was assessed via AFM, revealing surface roughness comparable to bulk Ge. While high-resolution transmission electron microscopy (HR-TEM), (Fig. 7c-ii) revealed that monocrystalline nature of Ge preserved from the Ge substrate into the grown Ge-epilayer without presence of dislocation or defects15; however, due to the limited field of view and sensitivity of HR-TEM, low-density defects below ~109/cm2 cannot be ruled out. The growth mechanism occurs in three different steps, as illustrated in Fig. 7c-iii46. Initially, the Ge nucleation on mesoporous Ge structure occurs on top surface of pore walls, forming nanoscale three-dimensional (3D) islands. Further growth of Ge increases the size of these islands, leading to coalescence and a temporary increase in surface roughness. Once a critical thickness (Tc), the minimum thickness required for the full islands’ coalescence into continuous dense membrane, was reached, the densified layer further thickened, and growth is dominated by 2D layer by layer growth mode. As a result, surface morphology is improved, and a decrease in surface roughness occurs. III-V epitaxy on Ge-epi Finally, to demonstrate that III–V growth on single mesoporous Ge layer + Ge-epi substrate can yield solar cell device performance comparable to that on bulk Ge substrates, a 2 µm-thick single-junction GaAs solar cell structure was grown by metal–organic chemical vapor deposition (MOCVD), as shown in Fig. 7c-iv, v80. A few dozen GaAs single-junction solar cells from different parts of the 100 mm Ge wafer (center and edge) were tested. The cells without antireflective coating demonstrated an average efficiency of ~22% under 1 sun at AM 1.5 G80. While this is below the state-of-the-art efficiency of GaAs solar cells ( ~ 29%)121, the reduced performance is primarily attributed to the absence of an antireflective coating, which limits mainly the short-circuit current density. The obtained VOC of 1.01 V suggests good material quality, although it remains below state-of-the-art values (~1.04–1.12 V)122,123, depending on the absorber type, with ~1.04 V for p-type and up to ~1.12 V for n-type GaAs)122. To sum up, each engineered manufacturing route has its own advantages and disadvantages. GeON method offers high manufacturing yield and great control and reproducibility, however processing steps and tools used are rather expensive. On the other hand, electrochemical etching routes are comparatively less expensive. The choice between Biporous mesoporous Ge and single mesoporous Ge layer + Ge-epi engineered substrates depends upon the application of the engineered Ge substrate; the Biporous mesoporous Ge approach is advantageous when substrate solely used for epi-template function. While single mesoporous Ge layer + Ge-epi approach is attractive when Ge is used also as an active junction in III-V with a thickness of about 10 µm. However, it adds an additional step of Ge epilayer growth on Ge porous layer, which can add up the total cost. Using the Biporous mesoporous Ge engineered substrate, the resulting solar cells exhibit bulk-like EQE and only a small VOC loss of only ~60 mV, which is attributed to surface defects79, whereas the single mesoporous Ge layer + Ge-epi shown comparable device efficiency for GaAs junction solar cells as of GaAs on bulk45. Solar cell fabrication This section discusses the different device architectures, as well as the transfer and microfabrication of solar cells after the deposition of III-V layers. Device architecture There are many different architectures of III-V solar cells exits, such as single junction solar cell (SJSC), 2JSC, 3JSC and so on, Fig. 8a. GaAs SJSC based on Ge substrate has shown the efficiency of about 21%80. MJSC consists of several subcells grown epitaxially on top of each other with increasing bandgap where each cell absorbed a different spectrum of the sunlight. As these subcells are connected in series so the current of MJSC is limited by the subcell which produces minimum photocurrent. One example is the three-junction solar cell (3JSC) with a GaInP/GaInAs/Ge configuration, grown in an upright lattice-matched (ULM) structure on a Ge substrate (Fig. 8a-ii)124.Fig. 8III-V solar cells architectures and controlled spalling technology for device detachment and transfer.a Schematic representation of different architecture of III-V solar cells (i) lattice-matched single junction solar cell, (ii) lattice matched 3JSC, (iii) upright metamorphic 4JSC with representation of the lattice constant variations along the growth direction, (iv) inverted metamorphic 4JSC with representation of the lattice constant variations along the growth direction, adopted. b Illustration of typical processing of III-V solar cells by controlled spalling technology, with (b) upright (double transfer) and (d) inverted device structure (single transfer)17. Digital image of as-fabricated III-V solar cells using controlled spalling technology with (c) upright and (e) inverted device structure. Devices in (c, e) were spalled manually by lifting the handle tapes attached on the Ni stressor layers. Panels (a, b, d) are reproduced from17, with permission from Elsevier. Panels (c, e) are reproduced from17; the original images were reproduced therein from130,132 John Wiley & Sons Ltd. and IOP Publishing Limited, respectively. Permissions for reuse of panels (c, e) were obtained from IOP Publishing Limited and John Wiley & Sons Ltd. through Copyright Clearance Center, Inc. III-V solar cells architectures and controlled spalling technology for device detachment and transfer. a Schematic representation of different architecture of III-V solar cells (i) lattice-matched single junction solar cell, (ii) lattice matched 3JSC, (iii) upright metamorphic 4JSC with representation of the lattice constant variations along the growth direction, (iv) inverted metamorphic 4JSC with representation of the lattice constant variations along the growth direction, adopted. b Illustration of typical processing of III-V solar cells by controlled spalling technology, with (b) upright (double transfer) and (d) inverted device structure (single transfer)17. Digital image of as-fabricated III-V solar cells using controlled spalling technology with (c) upright and (e) inverted device structure. Devices in (c, e) were spalled manually by lifting the handle tapes attached on the Ni stressor layers. Panels (a, b, d) are reproduced from17, with permission from Elsevier. Panels (c, e) are reproduced from17; the original images were reproduced therein from130,132 John Wiley & Sons Ltd. and IOP Publishing Limited, respectively. Permissions for reuse of panels (c, e) were obtained from IOP Publishing Limited and John Wiley & Sons Ltd. through Copyright Clearance Center, Inc. Although the efficiency of ULM 3JSC is close to the other III-V state of the art devices, the main challenge lies in the characteristics of Ge bottom cell, which must withstand the thermal load associated with the growth of upper subcell. This leads to a decrease in VOC of 55 mV, along with short current density (JSC) of 2.3 mA/cm2 in Ge subcell (due to thicker emitter resulting from deeper diffusion of III-V elements diffusion during thermal loading), compared to single-junction Ge cell, as reported and investigated by Solar Energy Institute of UPM (IES-UPM)125. Simulations results confirmed these losses and attributed them to degraded minority carrier properties in the emitter region, especially close to GaInP/Ge interface. To alleviate this thermal load, optimized nucleation strategies were implemented by reducing the thickness of the first two layers on top of Ge substrate (GaInP ( ~ 50 nm) and GaInAs ( ~ 500 nm)) and lowering the growth temperature of GaInAs from 675 °C to 600 °C, resulting in a recovery of up to ~50 mV in VOC in the complete 3JSC device compared to previous developed design by the group. For the details of full semiconductor structure of this ULM GaInP/GaInAs/Ge solar cell, reader is refereed to this ref. 125. Another challenge is that the Ge subcell typically generates ~50 % more current than the upper two subcells, which leads to excess current being dissipated as heat43. To compensate this excessive current in 3JSC a light harvester (InxGa1-xNyAs1-y) is added between Ge and InGaAs layers, which reduces light reaching Ge and hence the generated photocurrent in Ge sub cell. The bandgap and lattice constant of InxGa1-xNyAs1-y can be tuned by adapting In and B contents5,126. However, the addition of InxGa1-xNyAs1-y in lattice-matched configuration is challenging because common growth methods (metal organic phase epitaxy and molecular beam epitaxy) frequently produce electronic defects in the layer127. Therefore, new fabrication techniques developed to use lattice-mismatched materials. In these methods, the constraints of using lattice-matched semiconductor are overcome by depositing a metamorphic transparent buffer layer (MTBL) between the subcells128. These MTBLs gradually vary the lattice constant between successive layers, inevitably introducing misfit dislocations at the interface to accommodate lattice mismatch. Careful design of these layers minimizes the resulting threading dislocations, which can otherwise propagate into the active subcells. Depending on the growth direction (from narrow- to wide- bandgap subcells or vice versa) these are divided into upright metamorphic (UMM) and inverted metamorphic (IMM), Fig. 8a-iii and Fig. 8a-iv respectively. IMM configuration is advantageous because the two uppermost cells (wider-bandgap) grown on a lattice-matched substrate, followed by metamorphically growth of the remaining subcells, limiting defect propagation ( < 10−4 cm−2 vs. >10−6 cm−2 in UMM) and improving MJSC performance129. The diversity of engineered Ge substrate technologies offers significant opportunities for developing advanced solar cell architectures. For GeON, despite the need for costly infrastructure, its scalability, high crystalline quality, epi-readiness for III-V materials, and precise control over membrane thickness and detachability make it highly versatile. For instance, a few-micrometer-thick Ge membrane (without additional thickening) can serve solely as an epitaxial template for III-V growth, enabling structures such as 2 J ULM (without active Ge subcell), UMM, or IMM. When the Ge membrane is thickened by CVD, it can actively contribute to the device, supporting thin 3 J ULM or even multi-junction configurations like 4 J. Conversely, Ge membranes produced via the single mesoporous Ge layer + Ge-epi approach are relatively thin ( ≈ 1 µm) up to date, and thus primarily function as templates for III-V epitaxy, allowing the development of 2 J structures (ULM, UMM, or IMM) or higher-junction cells without an active Ge subcell. Future advancements in high-rate Ge epitaxy (e.g., CVD) could enable membrane thickening, facilitating solar cells with active Ge subcells. Finally, the Biporous route is mainly suited for architectures without an active Ge subcell, as Ge epitaxy is not part of this route, such as 2 J ULM, UMM or IMM. However, by incorporating a thick Ge epitaxial layer, this approach could also support structures with active Ge subcells, similar to the single mesoporous Ge layer + Ge-epi strategy. Transfer of solar cells device on definitive carrier Currently, there are two transfer routes of thin detachable solar cells reported in the literature. Through these transfer techniques, the detachable device is transferred onto a definitive carrier. The first one, simple-transfer process, applicable mainly to single-junction devices and inverted structures, and the second one, double transfer, for upright multijunction devices. The basic microfabrication process flow remains largely the same as for solar cells on bulk substrates, with the addition of supplementary steps required for handling the extremely thin membranes17. Simple transfer In the simple transfer approach, after the epitaxial growth of III-V layers on Ge substrate, and the back electrical contacts deposition on the top surface, the structure is bonded to a flexible carrier and detached from the substrate. In the following steps, the microfabrication process with front electrical contacts is completed on the exposed side. This method requires a single-junction configuration “inverted” epitaxial structure to ensure proper device functionality after transfer. This approach has been used in the framework of the spalling technique in order to transfer solar cell structure17,130. Finally, this approach has been widely used with the ELO process to transfer single-junction GaAs and inverted metamorphic triple-junctions131. Double transfer In the double transfer method, active III-V epilayers containing the sub-cells are first grown on a Ge substrate. A stressor layer is deposited on top of the structure followed by bonding to a temporary carrier. The active layers are released from the Ge substrate using a mechanical method. Back-side contacts are deposited, and the structure is bonded to a final carrier. Both the temporary carrier and stressor layer are then removed, exposing the front side, where microfabrication steps are carried out, including the formation of front contact and deposition of an anti-reflective coating17,132. Within the framework of porous lift-off approach, this approach allowed transfer of a single-junction GaAs solar cell81. It should be noted that several fabrication process-flow scenarios are possible, with the detachment step being a critical element that can be positioned at different stages of the process. For instance, fabrication steps such as front-side metallization may be performed either before or after detachment. The resulting process flow will depend on technological choices and infrastructure constraints. This fabrication scheme allows upright epi-structure configuration and does not require any modification of epitaxial growth or adaptation of epitaxial structure. Both simple and double transfer methods for the fabrication of solar cells are illustrated in the Fig. 8d, b respectively130,132, in the context of a spalling-based detachment approach; however, these process schemes are general and can be extended to other layer transfer techniques such as epitaxial lift-off (ELO)131. In the case of GeON-engineered substrates, the transfer of small-scale devices with a GaAs active layer onto inexpensive substrates such as glass has been demonstrated, achieving an efficiency of about 14%77. Similarly, the transfer of a complete solar-cell devices was demonstrated in another report; however, no device data is revealed on testing these detached devices119. For the Biporous mesoporous Ge engineered substrate, no device transfer demonstration have been reported, although detachment of biporous Ge layer itself from Ge wafer is demonstrated45. On a single mesoporous Ge layer + Ge-epi engineered substrate, recent work has demonstrated the detachment and transfer of functional GaAs solar-cell structures from parent Ge wafer using a stress-induced Ni-layer lift-off method; however, detailed photovoltaics parameters were not reported81. Detachment and adhesion strength tuning As discussed in the previous section, detachment of the Ge membrane from the parent substrate represents a crucial phase in the fabrication process, enabling further processing, integration, or reuse of the substrate. For the case of GeON, detachment of GeON foils was controlled by engineering the pillar density and geometry to balance adhesion and mechanical stability. Lithography defined hollow pillars with optimized pitch replaced randomly formed ones, enabling reproducible detachment without blistering or cracking. A relative detachment force scale was adopted, in which the reference pitch—defined as the lowest pitch in the lithography mask design—is labeled F₁and corresponds to the lowest detachment force. Pillar configurations corresponding to detachment force levels ≥ F200, where F200 = 200 × F1, demonstrated sufficient robustness for subsequent III–V epitaxy and device processing119. Building on these concepts, specifically in the context of the single mesoporous Ge layer + Ge-epi engineered substrate developed by our team at UdeS adopts a different strategy, where the adhesion strength of Ge epilayer grown on single mesoporous Ge layer can be tuned with the modification of porous layer characteristics (porosity and thickness) and annealing (growth temperature and/or post-annealing). Annealing during the growth of Ge epilayer or post-thermal annealing forms pillars that connect this Ge epilayer with Ge wafer. Consequently, the characteristics of the pillars (density and diameter) determine the adhesion strength of the grown Ge epilayer (Ge nanomembrane). Large pillar diameter and/or pillar density will increase the adhesion strength, and vice versa. The pillar characteristics could be described with one parameter (pillar coverage S), and given by the equation133:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S=d\Pi {\frac{D}{4}}^{2}$$\end{document}S=dΠD42 Where, d = pillar density and D = pillar diameter. Figure 9a shows that depending upon the range of annealing temperature (T), increasing T reduces the pillar surface coverage, while Fig. 9b shows that increasing pillar surface coverage gives high adhesion strength. The mesoporous Ge layer with high porosity tends to exhibit low adhesion strength to the Ge nanomembrane, since a large portion of the material is removed during porosification. Similarly, thicker mesoporous Ge layers show weaker adhesion because Ostwald ripening reduces the density (coverage) of pillars113. In the same way, the growth temperature has a crucial role in porous layer reorganization, inducing a tuning of the adhesion strength. For the single mesoporous Ge layer + Ge-epi engineered substrate, a buffer layer is first grown to protect the porous structure. During buffer-layer growth, increasing the temperature from 350 °C to 400 °C enlarges the surface pores, which allows incoming Ge atoms to fill them. This reduces the overall porosity and thereby increases the adhesion strength113. To further evaluate the adhesion strength a study shown that annealing produces non-linear effect on adhesion. As shown in the Fig. 9c two region can be distinguished, first, at the temperature (300 °C–700 °C) the pillars become smaller because atomic mobility within the mesoporous Ge layer increases just enough that Ge atoms tend to diffuse away from the pillar network into the overlying nanomembrane or into the underlying substrate, resulting in decrease in adhesion layer strength.Fig. 9Effect of annealing temperature on pillar surface coverage and adhesion strength of Ge membrane.a Pillar surface coverage variation with annealing temperature. b The adhesion strength variation with the pillar surface coverage, and (c) adhesion evolution as a function of annealing temperature; experimental value in red (average values with error bars minimum and maximum measurements), FEM simulation (filled area, minimum calculated with 40 MPa bulk fracture strength (fs), maximum calculated with 160 MPa bulk fracture strength). The best fit with experimental measurements is with a bulk fracture strength of 88 MPa (the black dashed lines are a guide for eyes) and the dashed zone indicates the temperatures with highly adhesive membrane (reprinted133, with permission from Elsevier) Effect of annealing temperature on pillar surface coverage and adhesion strength of Ge membrane. a Pillar surface coverage variation with annealing temperature. b The adhesion strength variation with the pillar surface coverage, and (c) adhesion evolution as a function of annealing temperature; experimental value in red (average values with error bars minimum and maximum measurements), FEM simulation (filled area, minimum calculated with 40 MPa bulk fracture strength (fs), maximum calculated with 160 MPa bulk fracture strength). The best fit with experimental measurements is with a bulk fracture strength of 88 MPa (the black dashed lines are a guide for eyes) and the dashed zone indicates the temperatures with highly adhesive membrane (reprinted133, with permission from Elsevier) To verify these experimental results, three-dimensional finite-element-method (FEM) simulations of membrane adhesion strength were performed using a semi-empirical model. The simulated adhesion strengths follow the same trend as the measurements, with the mean experimental adhesion strength lying between the highest (160 MPa) and lowest (40 MPa) simulated values. By fitting the FEM model to empirical data, a bulk Ge fracture strength of 88 MPa was determined, demonstrating that postgrowth membrane adhesion strength can be controllably tuned to achieve the desired detachment ease for Ge membrane133. Ge substrate reconditioning The final process after the detachment is the reconditioning of the parent substrate to reuse it for the next cycle. This reconditioning is critical to ensure that the parent substrates can be effectively reused, thereby achieving their cost effectiveness and sustainability. However, the substrate cannot be reused directly after detachment because of broken pillars and remaining surface features, which leads to high surface roughness. For the GeON engineered substrate, after the detachment of 1 µm thick Ge foil about 2 µm of rim is still present. Reconditioning of this substrate involves multiple steps from front side grinding, chemical etching for stress relief to front side polishing and cleaning, resulting in about 40 µm of loss in Ge thickness82. To minimise the loss in the thickness, polishing without grinding was tested, and despite the multiplication of passes and total removal of 40–50 µm Ge thickness, the recess could not be levelled82. However, this rim appears to be related to an engineering aspect and may be mitigated using a zero-rim mask. On the other hand, for the samples formed via electrochemical porosification, such as Biporous mesoporous Ge and single mesoporous Ge layer + Ge-epi engineered substrate, only chemical etching/wet etching was sufficient. Each reconditioning process removed about 5 µm from each side for a total of about 10 µm16. Wet etching is a cost-effective method based on a strategy similar to that developed for silicon (Si) without the need for the expensive conventional chemical-mechanical polishing (CMP) approach. Wet etching for Ge consists of two steps: (1) oxidation of surface Ge atoms by an oxidizing agent and (2) solubility of these oxidized compounds. In wet etching using hydrogen peroxide (H₂O₂), H₂O₂ acts as an oxidizing agent, converting surface Ge into Ge–O species that subsequently form the [Ge(OH)₂]²⁺ intermediate through surface Ge oxide134. However, using H2O2 30 wt% with pH = 3 results in a low etch rate of only 0.125 µm/min135,136. To increase the etch rate and make the process time compatible with industrial standards, complexing agents can be added to the oxidizing agent. This includes NH4OH in dilute Ammonia Peroxide Mixture (APM, NH4OH:H2O2:H2O, 1:1:5, v-v-v), HCl in dilute Hydrochloric acid Peroxide Mixture (HPM, HCl:H2O2:H2O, 1:1:7, v-v-v), or HF in HF-based solutions (HF:H2O2:H2O,1:3:6, v-v-v)137–139. The first demonstration of proof-of-concept of wet etching on the samples formed by electrochemical porosification for reconditioning to reuse is given by Chapotot et al16. A HF:H2O2:H2O solution (10:80:10, v/v/v) was used for reconditioning. Independent of pillar diameter on the substrate, ranging from 37 nm up to 500 nm and obtained after post-growth thermal annealing at various temperatures, smooth detached Ge surfaces were achieved with a very low surface roughness of approximately 2.5 nm. In addition, the Ge consumption was limited to only 10 μm in total (5 μm on each side). Moreover, it is claimed that this fast (4 min etch), cost-effective, and simple wet-etch reconditioning process, involving minimal Ge consumption, was reported to be compatible with industrial-scale requirements. Ge substrate reuse and sustainability The reuse of Ge substrate certainly has both an economic and an environmental impact. As mentioned in the introduction the Ge substrate can be reused after detachment. The single mesoporous Ge layer + Ge-epi approach, developed by UdeS team for the fabrication of Ge membranes, consists of several steps from porosification to reconditioning15, and has demonstrated up to three reuses of the Ge substrate, thereby offering the potential to significantly reduce Ge consumption78. Reducing Ge consumption is crucial because it is a critical raw material due to its economic importance, supply risk and scarcity41. For a conventional 100 mm Ge wafer with a thickness of 145 µm and without substrate reuse (Fig. 10), approximately 6.06 g of Ge is passed onto device processing, e.g., solar-cell fabrication, of which about 2.72 g is typically lost during thinning to 80 µm, although much of this material could potentially be recovered through recycling. By contrast, the single mesoporous Ge layer + Ge-epi approach has been shown to require only about 0.27 g of Ge per cycle (0.21 g for the 5 µm epitaxial active layer plus 0.06 g Ge lost during wastewater recycling, which can also be recycled). Compared with the 2.72 g lost in conventional thinning, this corresponds to an ≈90% reduction in Ge consumption, while relative to the total 6.06 g initially passed on device processing, it represents an ≈95% reduction, highlighting the potential of single mesoporous Ge layer + Ge-epi method15. Moreover, it is claimed that 600 µm-thick Ge wafers can be reused multiple times, reducing Ge consumption, carbon emission and cost; while 30 reuses have been suggested in principle, current report demonstrate up to three reuses only78. Importantly, although reducing Ge consumption is a necessary step toward improved resource efficiency, it does not by itself render space technologies sustainable, given the broader environmental impacts associated with launch emissions and orbital debris7,8. In addition, potential rebound effects must be considered: reductions in material use and cost may accelerate the deployment of large-scale satellite constellations, potentially increasing the total consumption of Ge and III–V materials. Therefore, the benefits of the single mesoporous Ge layer + Ge-epi approach should be viewed within a broader life-cycle perspective, contributing to but not solely ensuring improved sustainability in space applications.Fig. 10Comparison of Ge consumption in conventional and mesoporous Ge-based solar cell fabrication.Comparison of Ge consumption between (a) the Ge wafer pathway and (b) the single mesoporous Ge layer + Ge-epi approach in the case of solar fabrication thickness (reproduced from15, with permission of Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.). Comparison of Ge consumption in conventional and mesoporous Ge-based solar cell fabrication. Comparison of Ge consumption between (a) the Ge wafer pathway and (b) the single mesoporous Ge layer + Ge-epi approach in the case of solar fabrication thickness (reproduced from15, with permission of Royal Society of Chemistry, permission conveyed through Copyright Clearance Center, Inc.). Toward commercialization: recommendations and improvements For the Ge wafer reuse strategies to transition from laboratory demonstration to industrial production, several focused optimizations are required. GeON engineered substrates demonstrated high quality, up to 200 nm detachable membranes with high-performance materials but demand capital-intensive processing infrastructure. In addition, the wafer reconditioning process was demonstrated with a loss of about 40 µm of Ge thickness per cycle. Developing modular, cost-efficient equipment and reducing material loss during reconditioning could potentially reduce capital requirements, minimize Ge consumption, and improve scalability. Nevertheless, the GeON approach remains very appealing for high-end products, where the specialized tools required for GeON fabrication are often already part of the device processing infrastructure, minimizing additional equipment investment. On the other hand, for the electrochemical porosification routes, electrochemical porosification has reached a high level of control in the recent years but its slow rate limits throughput. Incorporating automation and optimizing electrochemical parameters can reduce porosification time from ~30 min toward sub-10 min regimes, improving throughput while requiring careful control to maintain pore morphology and uniformity. Lastly, while single mesoporous Ge layer + Ge-epi engineered substrates use Molecular Beam Epitaxy (MBE) for the growth of Ge epi, this method, though precise, is inherently slow and not industrially scalable; transitioning to high-throughput epitaxial growth techniques such as Chemical Vapor Deposition (CVD) and Hydride Vapor Phase Epitaxy (HVPE) would enable higher material throughput and improve compatibility with existing semiconductor manufacturing infrastructure140. In addition, there are ongoing efforts to adopt screen-printing metallization methods, originally developed and validated for silicon photovoltaic processing, to III-V device fabrication, with the aim of reducing manufacturing costs and increasing production throughput141. Finally, although small-scale fabrication of complete solar cell structures and their successful detachment and transfer have been demonstrated, wafer-scale, defect-free lift-off remains a significant challenge. The quality and uniformity of released films with integrated device at the full-wafer scale are not yet well characterized. Nevertheless, the development of stress-release layers and robust mechanical handling methods could, in the future, enable complete and reproducible lift-off, thereby improving yield and device performance. Collectively, implementing these strategies will accelerate the industrial adoption of Ge wafer-reuse approaches, potentially reducing Ge consumption15,45 and thereby lowering carbon emissions and production costs, while enabling high-efficiency III-V/Ge multijunction solar cells for space applications. Conclusions /outlook This review summarizes recent advances in germanium membrane technology as a pathway to lightweight, cost-efficient templates for III-V multijunction solar cells. Among three approaches for fabrication of Ge membrane using porosification-based lift-off and substrate reuse, GeON engineered substrates provide the highest yield but require a more expensive infrastructure, whereas electrochemical porosification routes methods (Biporous mesoporous Ge and single mesoporous Ge layer + Ge-epi) are potentially more affordable yet limited by long processing times and low throughput. For single mesoporous Ge layer + Ge-epi engineered substrates, epitaxial growth of Ge via MBE remains a bottleneck due to its slow growth rate and limited scalability. In addition to that, although small-scale detachment of complete solar cell structures has been demonstrated, reproducible wafer-scale liftoff is still a key challenge. Continued progress in process optimization, scalable epitaxy, and mechanical handling is essential for the industrial realization of Ge-membrane-enabled multijunction solar cells. Looking ahead, further development of epitaxial technologies for Ge growth will be essential to unlock the full potential of Ge membranes in high-volume manufacturing. While growth rate is an important figure of merit and can be improved by switching from MBE to CVD for Ge-epi growth, overall throughput is fundamentally constrained by temperature ramp-up and cool-down cycles as well as auxiliary process steps in CVD. More importantly, even if single mesoporous Ge layer + Ge-epi or substrate reuse is optimized, this will not directly translate to higher production volumes of Ge based III-V MJSCs because III-V epitaxy, particularly the lengthy process steps in MOCVD and HVPE remains the dominant bottleneck for scaling production, although recent developments in dynamic HVPE (D-HVPE) have demonstrated substantially increased III–V epitaxy throughput (growth rates ≳300 µm h−1) and promising device performance142,143. Similarly, high-growth-rate metal-organic vapor phase epitaxy (MOVPE) has demonstrated GaAs growth rates exceeding ~280 µm h−1, while maintaining excellent material quality and high solar cell open-circuit voltages even at growth rates of ~100 µm h−1123. Beyond photovoltaics and thermophotovoltaics, Ge membranes can offer a transformative platform for next-generation electronics and photonics53. Their compatibility with back-end-of-line (BEOL) thermal budgets, typically limited to 400–450 °C, can enable direct integration with CMOS circuitry while preserving existing metallization and interconnect reliability, thus overcoming the 600–800 °C temperature barrier of conventional heteroepitaxy54. Furthermore, Ge membranes can serve as versatile templates for the growth of III–V and III-nitride semiconductors, enabling the fabrication of high-performance photonic and optoelectronic devices. Their mechanical flexibility can allow transfer onto diverse substrates, including polymers, PET, and curved surfaces, paving the way for flexible displays, wearable electronics, and multifunctional chip architectures144. Sustainability considerations, including substrate reuse and low-temperature processing, further enhance the appeal of Ge membrane technologies. However, the overall economic feasibility remains sensitive to equipment costs, process yield, and integration complexity. A coherent technology development roadmap is emerging, outlining the current state of readiness, key technical challenges, and future directions. This includes efforts to optimize porosification kinetics, improve mechanical robustness during transfer, and validate large-scale integration strategies. With continued interdisciplinary innovation, Ge membranes are poised to play a central role in next-generation flexible electronics, high-efficiency solar energy systems, and integrated photonic platforms, bridging the gap between fundamental research and scalable, real-world applications. Overall, fabrication of Ge membranes and Ge-wafer-reuse can serve as an important building block for heterojunction device platforms while reusing of Ge wafer preserves the critical raw material, reduces the cost and enables the sustainable use of Ge. Supplementary information Transparent Peer Review file Transparent Peer Review file