No abstract

We have investigated the effect of reducing the thickness of strontium bismuth tantalate film to as low as 25 nm on its ferroelectric characteristics. A degradation of ferroelectric properties such as significant reduction in remanent polarization is generally observed with reduction in film thickness, in particular below 100 nm. This has been overcome by using a modified deposition process sequence and a crystallization technique based completely on the rapid thermal annealing process. The resulting ultrathin films show good remanent polarization, low-voltage saturation, low leakage current, high breakdown strength, and good endurance. These films demonstrate the potential for scaling and are excellent candidates for several generations of ferroelectric random access memory applications.

No abstract

Wafer bonding can be substituted for heteroepitaxy when manufacturing specific heterojunction-based devices. Devices manufactured using wafer bonding include multijunction solar cells, integrated sensors, heterogeneously integrated photonic devices on Si (such as high-performance laser diodes), Mach-Zehnder modulators, photodetectors, optical filters, and surface acoustic wave devices. In these devices, creating heterointerfaces between different semiconductors with heavily mismatched lattice constants and/or significant thermal expansion mismatch presents significant challenges for heteroepitaxial growth. High costs and poor yields in heavily mismatched heteroepitaxy can be addressed by wafer bonding in these optoelectronic devices and sensors, including the LiTaO3/Si and LiTaO3/SiO2 heterostructures. In the present work, heterostructure formation between piezoelectric LiTaO3 (100) and Si (100) and α-quartz SiO2 (100) is investigated via wafer bonding. Direct bonding is selected instead of heteroepitaxy due to a significant thermal expansion mismatch between LiTaO3 and Si-based materials. The coefficient of thermal expansion (CTE) of LiTaO3 is 18.3 × 10−6/K. This is 1 order of magnitude larger than the CTE for Si, 2.6–2.77 × 10−6/K and 25–30 times larger than the CTE for fused SiO2 and quartz (which ranges 0.54–0.76 × 10−6/K). Thus, even at 200 °C, a 4 in. LiTaO3/Si bonded pair would delaminate with LiTaO3 expanding 300 μm in length while Si would expand only by 40 μm. Therefore, direct wafer bonding of LiTaO3/Si and LiTaO3/SiO2 is investigated with low temperature (T < 500 K) Nano-Bonding™, which uses surface energy engineering (SEE). SEE is guided by fast, high statistics surface energy measurements using three liquid contact angle analysis, the van Oss/van Oss–Chaudhury–Good theory, and a new, fast Drop Reflection Operative Program analysis algorithm. Bonding hydrophobic LiTaO3 to hydrophilic Si or SiO2 is found to be more effective than hydrophilic LiTaO3 to hydrophobic Si or SiO2 temperatures for processing LiTaO3 are limited by thermal decomposition LiTaO3 into Ta2O5 at T ≥ 180 °C due to Li out-diffusion as much as by LiTaO3 fractures due to thermal mismatch.

The effects of crystal orientation and doping on the surface energy, γT, of native oxides of Si(100) and Si(111) are measured via Three Liquid Contact Angle Analysis (3LCAA) to extract γT, while Ion Beam Analysis (IBA) is used to detect Oxygen. During 3LCAA, contact angles for three liquids are measured with photographs via the “Drop and Reflection Operative Program (DROP™). DROP™ removes subjectivity in image analysis, and yields reproducible contact angles within < ±1°. Unlike to the Sessile Drop Method, DROP can yield relative errors < 3% on sets of 20-30 drops. Native oxides on 5 x 1013 B/cm3 p- doped Si(100) wafers, as received in sealed, 25 wafer teflon boats continuously stored in Class 100/ISO 5 conditions at 24.5°C in 25% controlled humidity, are found to be hydrophilic. Their γT, 52.5 ± 1.5 mJ/m2, is reproducible between four boats from three sources, and 9% greater than γT of native oxides on n- doped Si(111), which averages 48.1 ± 1.6 mJ/m2 on four 4” Si(111) wafers. IBA combining 16O nuclear resonance with channeling detects 30% more oxygen on native oxides of Si(111) than Si(100). While γT should increase on thinner, more defective oxides, Lifshitz-Van der Waals interactions γLW on native oxides of Si(100) remain at 36 ± 0.4 mJ/m2, equal to γLW on Si(111), 36 ± 0.6 mJ/m2, since γLW arises from the same SiO2 molecules. Native oxides on 4.5 x 1018 B/cm3 p+ doped Si(100) yield a γT of 39 ± 1 mJ/m2, as they are thicker per IBA. In summary, 3LCAA and IBA can detect reproducibly and accurately, within a few %, changes in the surface energy of native oxides due to thickness and surface composition arising from doping or crystal structure, if conducted in well controlled clean room conditions for measurements and storage.