Crystallization from an amorphous precursor presents a new route to control the properties of complex oxides by selecting their nanoscale morphology. A key challenge in crystal growth from the amorphous form is to select the locations of nucleation and the crystallographic orientation of the resulting crystals. Nucleation sites for crystallization of the prototypical perovskite complex oxide SrTiO3 (STO) from its amorphous form can be reproducibly introduced using nanoscale seed crystals. The results of two seeding strategies are reported here: (i) SrRuO3 (SRO) (001)-oriented nanomembranes and (ii) STO nanocrystalline seeds. Amorphous STO crystallizes laterally over distances of several microns from the seeds before encountering separately nucleated crystals. The lateral crystallization rates for both types of seeding methods are close to the values measured in solidphase epitaxy (SPE) of STO on single-crystal substrates. The lateral crystallization distances along different crystallographic orientations are equal, indicating that the lateral crystallization rate is isotropic. The isotropic crystallization suggests that the rate-limiting steps for crystallization occur within the amorphous STO away from the amorphous/crystalline interface. In addition to the lateral crystallization, STO crystallizes on top of the planar SRO nanomembrane seeds via SPE, forming partially relaxed [001]-oriented heteroepitaxial STO layers. The in-plane orientation of laterally crystallized STO near the SRO nanomembrane seeds exhibits overall polycrystallinity with regions in which micron-scale grains of laterally crystallized STO share the same in-plane orientation as the SRO nanomembrane seeds. Crystallization from seed crystals provides opportunities to create a wide range of other perovskite oxides in nanoscale geometries.

Methods to integrate different crystal orientations, strain states, and compositions of semiconductors in planar and preferably flexible configurations may enable nontraditional sensing-, stimulating-, or communication-device applications. We combine crystalline-silicon nanomembranes, patterning, membrane transfer, and epitaxial growth to demonstrate planar arrays of different orientations and strain states of Si in a single membrane, which is then readily transferable to other substrates, including flexible supports. As examples, regions of Si(001) and Si(110) or strained Si(110) are combined to form a multicomponent, single substrate with high-quality narrow interfaces. We perform extensive structural characterization of all interfaces and measure charge-carrier mobilities in different regions of a 2D quilt. The method is readily extendable to include varying compositions or different classes of materials.

To assess possible improvements in the electronic performance of two-dimensional electron gases (2DEGs) in silicon, SiGe/Si/SiGe heterostructures are grown on fully elastically relaxed single-crystal SiGe nanomembranes produced through a strain engineering approach. This procedure eliminates the formation of dislocations in the heterostructure. Top-gated Hall bar devices are fabricated to enable magnetoresistivity and Hall effect measurements. Both Shubnikov-de Haas oscillations and the quantum Hall effect are observed at low temperatures, demonstrating the formation of high-quality 2DEGs. Values of charge carrier mobility as a function of carrier density extracted from these measurements are at least as high or higher than those obtained from companion measurements made on heterostructures grown on conventional strain graded substrates. In all samples, impurity scattering appears to limit the mobility.

The field of oxide electronics has benefited from the wide spectrum of functionalities available to the ABO3 perovskites, and researchers are now employing defect engineering in single crystalline heterostructures to tailor properties. However, bulk oxide single crystals are not conducive to many types of applications, particularly those requiring mechanical flexibility. Here, we demonstrate the realization of an all-oxide, single-crystalline nanomembrane heterostructure. With a surface-to-volume ratio of 2 × 10(7), the nanomembranes are fully flexible and can be readily transferred to other materials for handling purposes or for new materials integration schemes. Using in situ synchrotron X-ray scattering, we find that the nanomembranes can bond to other host substrates near room temperature and demonstrate coupling between surface reactivity and electromechanical properties in ferroelectric nanomembrane systems. The synthesis technique described here represents a significant advancement in materials integration and provides a new platform for the development of flexible oxide electronics.

We report the imaging of nanoscale distributions of lattice strain and rotation in complementary components of lithographically engineered epitaxial thin film semiconductor heterostructures using synchrotron x-ray Bragg projection ptychography (BPP). We introduce a new analysis method that enables lattice rotation and out-of-plane strain to be determined independently from a single BPP phase reconstruction, and we apply it to two laterally adjacent, multiaxially stressed materials in a prototype channel device. These results quantitatively agree with mechanical modeling and demonstrate the ability of BPP to map out-of-plane lattice dilatation, a parameter critical to the performance of electronic materials.

Thin-film deposition on ultra-thin substrates poses unique challenges because of the potential for a dynamic response to the film stress during deposition. While theoretical studies have investigated film stress related changes in bulk substrates, little has been done to learn how stress might evolve in a film growing on a compliant substrate. We use silicon nanomembranes (SiNMs), extremely thin sheets of single-crystalline Si, as a substrate for the growth of amorphous SiNx to begin to address this question. Nanomembranes are released from a silicon-on-insulator wafer with selective etching, transferred over a hole etched into a Si wafer, and bonded to the edges of the hole. The nanomembrane window provides the substrate for SiNx deposition and a platform, using Raman spectroscopy, for measurements of the evolving strain in the nanomembrane. From the strain in the nanomembrane, the film stress can be inferred from the required balance of forces in the film/substrate system. We observe that the strain in the tethered NM increases as the NM is made thinner while the intrinsic steady-state stress in the deposited film is reduced.

Fast flexible electronics operating at radio frequencies (>1 GHz) are more attractive than traditional flexible electronics because of their versatile capabilities, dramatic power savings when operating at reduced speed and broader spectrum of applications. Transferrable single-crystalline Si nanomembranes (SiNMs) are preferred to other materials for flexible electronics owing to their unique advantages. Further improvement of Si-based device speed implies significant technical and economic advantages. While the mobility of bulk Si can be enhanced using strain techniques, implementing these techniques into transferrable single-crystalline SiNMs has been challenging and not demonstrated. The past approach presents severe challenges to achieve effective doping and desired material topology. Here we demonstrate the combination of strained- NM-compatible doping techniques with self-sustained-strain sharing by applying a strain-sharing scheme between Si and SiGe multiple epitaxial layers, to create strained print-transferrable SiNMs. We demonstrate a new speed record of Si-based flexible electronics without using aggressively scaled critical device dimensions.

We demonstrate microwave flexible thin-film transistors (TFTs) using transferrable single crystalline Si nanomembranes (NMs). By combining self-sustained strain techniques and strain compatible doping techniques, we realize a new speed record, fmax of 15 GHz, for flexible microwave TFTs based on Si. The TFTs exhibit superior mechanical flexibility. The demonstrations showed the great potential to further develop flexible microwave components and systems.

Silicon, germanium, and related alloys, which provide the leading materials platform of electronics, are extremely inefficient light emitters because of the indirect nature of their fundamental energy bandgap. This basic materials property has so far hindered the development of group-IV photonic active devices, including diode lasers, thereby significantly limiting our ability to integrate electronic and photonic functionalities at the chip level. Here we show that Ge nanomembranes (i.e., single-crystal sheets no more than a few tens of nanometers thick) can be used to overcome this materials limitation. Theoretical studies have predicted that tensile strain in Ge lowers the direct energy bandgap relative to the indirect one. We demonstrate that mechanically stressed nanomembranes allow for the introduction of sufficient biaxial tensile strain to transform Ge into a direct-bandgap material with strongly enhanced light-emission efficiency, capable of supporting population inversion as required for providing optical gain.

We demonstrate the feasibility of fabricating heterojunctions of semiconductors with high mismatches in lattice constant and coefficient of thermal expansion by employing nanomembrane bonding. We investigate the structure of and electrical transport across the interface of a Si/Ge bilayer formed by direct, low-temperature hydrophobic bonding of a 200 nm thick monocrystalline Si(001) membrane to a bulk Ge(001) wafer. The membrane bond has an extremely high quality, with an interfacial region of ∼1 nm. No fracture or delamination is observed for temperature changes greater than 350 °C, despite the approximately 2:1 ratio of thermal-expansion coefficients. Both the Si and the Ge maintain a high degree of crystallinity. The junction is highly conductive. The nonlinear transport behavior is fit with a tunneling model, and the bonding behavior is explained with nanomembrane mechanics.

Strain in a material changes the lattice constant and thereby creates a material with new properties relative to the unstrained, but chemically identical, material. The ability to alter the strain (its magnitude, direction, extent, periodicity, symmetry, and nature) allows tunability of these new properties. A recent development, crystalline nanomembranes, offers a powerful platform for using and tuning strain to create materials that have unique properties, not achievable in bulk materials or with conventional processes. Nanomembranes, because of their thinness, enable elastic strain sharing, a process that introduces large amounts of strain and unique strain distributions in single-crystal materials, without exposing the material to the formation of extended defects. We provide here prescriptions for making new strained materials using crystal symmetry as the driver: we calculate the strain distributions in flat nanomembranes for two-fold and four-fold elastically symmetric materials. We show that we can controllably tune the amount of strain and the asymmetry of the strain distribution in elastically isotropic and anisotropic materials uniformly over large areas. We perform the experimental demonstration with a trilayer Si(110)/Si((1-x))Ge(x)(110)/Si(110) nanomembrane: an elastically two-fold symmetric system in which we can transfer strain that is biaxially isotropic. We are thus able to make uniformly strained materials that cannot be made any other way.

Many important materials cannot be grown as single crystals in bulk form because strain destroys long-range crystallinity. Among them, alloys of group IV semiconductors, specifically SiGe alloys, have significant technological value. Using nanomembrane strain engineering methods, we demonstrate the fabrication of fully elastically relaxed Si1-xGex nanomembranes (NMs) for use as growth substrates for new materials. To do so, we grow defect-free, uniformly and elastically strained SiGe layers on Si substrates and release the SiGe layers to allow them to relax this strain completely as free-standing NMs. These SiGe NMs are transferred to new hosts and bonded there. We confirm the high structural quality of these new materials and demonstrate their use as substrates for technologically relevant epitaxial films by growing strained-Si layers and thick, lattice-matched SiGe alloy layers on them.

Defect-free, smooth, tensilely strained Si(110) has been fabricated via elastic strain relaxation of trilayer Si/SiGe/Si(110) nanomembrane heterostructures grown on silicon-on-insulator (SOI), in which the middle layer is compressively strained. We release these trilayer structures via selective etching to allow elastic strain sharing between the Si/SiGe/Si(110) layers, thus creating tensilely strained Si(110). We use low-temperature molecular beam epitaxy (MBE) to grow the SiGe and top Si layers to create smooth films and incorporate more strain in the Si(110) layers; the low growth temperatures suppress relaxation, allowing for thicker SiGe(110) films. The strain in the Si(110) layers can be controlled with elastic strain engineering up to ~0.7%.

Semiconductor nanomembranes, single-crystal sheets as thin as ten nanometers, offer many opportunities for novel devices and new science. The most interesting involve epitaxy to introduce strain at both local and global levels. Coming into play are membrane thinness, access to both sides of a sheet, transferability, and enhanced compliancy. Advances in Group IV optoelectronics, thermoelectrics, and photonics may be achievable by combining epitaxy with Si and Ge nanomembranes. Nanoepitaxy allows formation of new strained materials, periodic strain lattices, and mix and match membranes with hybrid orientations or compositions.

The desire for increased processor speed leads to a demand for high-carrier-mobility CMOS devices. The complementary nature of CMOS, with both n-type and p-type channels, means that the lowest-mobility channel will limit the device speed. In the conventional (001) orientation of Si. the hole mobility is dramatically less than the electron mobility (1), (2), and hence serves as a bottleneck in CMOS performance. To compensate for the lower hole mobility, it is customary to fabricate the p-type device regions 3-10x larger than the n-type regions, consuming an undesirably large quantity of device real. estate. The Current drive imbalance between n-type and p-type channels can be minimized, thus negating the need for disproportionately large p-type regions, by fabricating mixed regions of Si(110) (high hole mobility) and Si(001) (high electron mobility) on a single substrate; so-called hybrid-orientation technology (HOT) (1). We fabricate a mixed-crystal-orientation material in flexible membrane form, using Si nanomembrane (SiNM) transfer and overgrowth, to produce a "quilt" of Si(001) and Si(110).