Implantation, Wafer Bonding & Layer Transfer
Abstract
|
S. Christiansen, O. Moutanabbir, and V. Schmidt
Group-III-Nitride materials have been known up to now almost exclusively for opto-electronic applications. However, todays engineered substrates based on layer stacks, which can be obtained by implantation of light elements such as He or H, wafer bonding and layer transfer, allow for novel combinations of properties and thus design of novel devices (e.g. in the fields of high power and high frequency electronics) with performances that were not in reach with any kind of bulk substrates. For example, radio-frequency devices have been built so far exclusively on bulk materials such as GaAs, InP and SiC. Better performance of these devices in terms of operating voltage, maximum power or operating temperature ask for substrates that are different from the best bulk material of known type. I.e. custom made, engineered substrates appear to become one of the major limiting factors in the design of novel devices.
GaN has been in the past years demonstrated as the material of choice for High Power Radio Frequency High Electron Mobility Transistors ( HEMT) . But high crystalline quality single crystal GaN epitaxial layers can only be grown on SiC or GaN substrates (both being expensive and limited in wafer size and quantity). Alternatively, sapphire and bulk silicon can be used as substrates for GaN HEMT, but device performance on these wafers is clearly reduced due to poor thermal conductivity.
Based on wafer bonding and layer transfer a cost efficient solution to engineered substrates may be in reach that will leverage the use of advanced high power devices in wireless communication systems.
Development at both ends is needed: optimize the layer transfer process and the growth of large diameter, high quality free standing wafers of group-III nitrides, ZnO, SiC.
Fig. 1: Schematic of the process flow of the “smart-cutR” process: 1) Implantation of light ions below the wafer surface 2) Direct wafer bonding 3) Annealing of the implanted wafer leads to splitting parallel to the wafer surface due to agglomeration of the implanted ions and thereby crack formation 4) Planarizing of the split wafer surface by polishing or etching.
Fig. 2: 1) Schematic of the blistering (1,3) and splitting (2,4) process after high dose H+ implant. (1) and (2) show as implanted wafers and (3) and (4) show wafers after annealing when the over pressure increases and yields blistering/splitting.
|
|
Fig. 3: Implanted and annealed wafer: a) Cross-sectional TEM micrograph (multi-beam conditions; electron beam parallel to the (110)-pole) showing the exfoliation of a thin layer of GaN by stable crack growth parallel to the layer surface (which is not seen in the micrograph); b) optical micrograph showing the blistering on the wafer surface; open and closed blisters of ~ 10 µm are visible.
High quality & inexpensive substrates of wide bandgap materials for photonic and electronic applications based on wafer bonding and layer transfer
Right now status and perspectives of ZnO, SiC and III-nitrides bulk or engineered substrates is to be defined. Perspectives depend on bulk crystal growth and epitaxy achievements which are related to epi-ready surface preparation, for applications optimized conductivity range, tolerable dislocation densities, influence of residual lattice strain etc.
GaN: HVPE is nowadays an established method to produce free-standing 2” GaN substrates. However, these wafers are very expensive and, being the result of hetero-epitaxial growth, still contain a high dislocation density, which makes them unsuitable for laser fabrication. A further drawback is the wafer bending after separation from the original substrate. Alternative growth methods are under investigation such as high-pressure solution growth, ammono-thermal growth, physical vapor transport etc. have been tried. A breakthrough would be the production of engineered substrates based on implantation, waferbonding and layer transfer.
AlN: The thermodynamic properties of AlN seem to suggest sublimation as the most promising technique for its bulk growth. Single crystal boules of 1 inch diameter have already been demonstrated and 2 inch ingots are presently under development. Are there alternatives to AlN sublimation and is implantation, wafer bonding and layer transfer an option?
SiC: In this case 3” substrates are already available. The main obstacles are to overcome nano/micropipes and polytype instabilities and to reduce dislocation densities and stacking faults.
ZnO: This material is undergoing a renaissance. High-quality bulk substrates and solving unresolved doping difficulties would allow the development of novel or better opto-electronic devices. However, so far only tiny substrates are available for epitaxial growth. In contrast to nitrides and SiC, ZnO could in principle be grown from the melt as the melting point and dissociation pressure are compatible with todays furnaces. Again is implantation, wafer bonding and layer transfer an option for cheaper high quality substrates with tailored properties?
Fig. 1: (left) AFM image of a ZnO surface after exfoliation; (right) XTEM of an exfoliated layer; surface roughness and implantation induced voids are discernible.