Gallium nitride ͑GaN͒ thin films grown on sapphire substrates were successfully bonded and transferred onto Si substrates using a Pd-In metallic bond. After bonding, a single 600 mJ/cm 2 , 38 ns KrF ͑248 nm͒ excimer laser pulse was directed through the transparent sapphire followed by a low-temperature heat treatment to remove the substrate. Channeling-Rutherford backscattering measurements revealed the thickness of the defective interfacial region to be approximately 350 nm. The full width at half maximum, low-temperature ͑4 K͒, donor-bound exciton photoluminescence ͑PL͒ peak was larger by 25% on the exposed interfacial layer compared to the original GaN surface. Ion milling of the exposed interface to a depth of 400 nm was found to remove the interfacial layer and associated defects. The minimum channeling yield and PL linewidths from the exposed interface were found to be comparable to those obtained from the original GaN surface after ion milling.

No abstract

The technique of transferring patterned ion-cut layers from one Si wafer to another was demonstrated. The starting silicon wafer was masked with checkerboard and line patterns with a 3 m thick polymethylmethacrylate/photoresist and was implanted with 5ϫ10 16 H ϩ ions/cm 2 at 150 keV. After stripping off the mask, the wafer was bonded to an oxide-coated receptor wafer through low-temperature direct wafer bonding. Heat treatment of this bonded pair showed that the hydrogen-induced silicon surface layer cleavage ͑ion cut͒ could propagate throughout about 16 mϫ16 m of nonimplanted material with implanted regions only 4 m wide. Mask width, spacing, and implantation profiles through the mask shape were shown to have effects on the internal microfracturing mechanisms. © 1998 American Institute of Physics. ͓S0003-6951͑98͒00645-7͔Three-dimensional electronic device integration offers significant opportunities for future system improvement in microprocessors and memories. 1,2 This prospect might be implemented with the hydrogen-induced silicon layer cleavage process, which has already been reported for capacitor patterns ͑passive devices͒. 3 To cleave the implanted layer, a minimum dose of a few times 10 16 /cm 2 of implanted hydrogen is needed. 4 This large dose of hydrogen most likely will damage the devices fabricated on the silicon prior to the ion-cut process. In this study, we introduce a patterned ioncut process in which active regions of the wafer are protected from the hydrogen implantation.In this study, Czochralski-grown, ͑100͒, n-type ( ϭ5 -50 ⍀ cm), 100 mm silicon wafers were used. The Si donor wafer was coated with a layer of KTI 950K 9% polymethylmethacrylate ͑PMMA͒ and a layer of Shipley 1400-30 photoresist with a total thickness of 3 m, followed by patterning of various sizes of squares and lines for the implantation mask with different openings for the hydrogen implantation ͑see Fig. 1͒. This patterned wafer was then implanted with H ϩ ions at 150 keV with a dose of 5ϫ10 16 cm Ϫ2 . Dur-ing implantation, the wafer was kept at ambient temperature. A 3 m thick ion mask layer ͑including PMMA and the photoresist͒ was applied to prevent the hydrogen ions from reaching the silicon wafer surface, resulting in hydrogen-ion implantation only in the openings. After the implantation, the ion mask was removed by oxygen plasma ashing. To determine the cracking temperature, the patterns were annealed at temperatures from 400 to 600°C in forming gas for 5 min after the removal of the ion mask. It was found that blistering occurred between 500 and 550°C. It was clear under the microscope that all the blisters were confined to the implanted regions. This observation confirms the effectiveness of the implant mask for the protected regions.On the receptor wafer, a layer of thermal oxide 200 nm thick was grown. The two wafers were bonded directly faceto-face at room temperature or at slightly elevated temperature after standard RCA cleaning of the implanted wafer. The bonded pair was then heated in a rapid thermal annealer un...

The ion source for the 1 MW National Spallation Neutron Source (NSNS) is required to provide 35 mA of H- beam current (1 ms pulses at 60 Hz) at 65 keV with a normalized root-mean-square emittance of <0.2 pi mm mrad. The same ion source should be able to produce 70 mA of H- at 6% duty factor when the NSNS is upgraded to 2 MW of power. For this application, a radio-frequency driven, magnetically filtered multicusp source is being developed at Lawrence Berkeley National Laboratory. The design of this R and D ion source, which is equipped with a cesium dispenser-collar, a fast ion beam prechopper (rise times <100 ns) and a strong permanent-magnet insert for electron deflection, will be presented.

The upgrade of the Los Alamos Neutron Science Center (LANSCE) will require an ion source producing high intensity H− beams. The new LANSCE source in particular will need to generate 40 mA of H− beam current at a duty factor of 12% (1 ms pulse at 120 Hz). To achieve this, the present Los Alamos results were first reproduced employing a prototype surface conversion source similar to the existing LANSCE source. Using these results as a benchmark for further testing, it was discovered that by moving the filament cathode into the chamber’s cusp-field, the H− ion yield was enhanced. Considering this result, two extension chambers were added with movable magnetic filters at each end of the source to further improve the H− beam current. In addition, construction has been started on a new prototype axial source which will enable more plasma ions to funnel across the filter field into the central region where the converter is located. Results of the magnetic filter operation as well as the new axial source design will be presented.