Show simple item record

dc.contributor.advisorRubloff, Gary Wen_US
dc.contributor.authorLei, Weien_US
dc.date.accessioned2006-06-14T05:35:06Z
dc.date.available2006-06-14T05:35:06Z
dc.date.issued2006-03-16en_US
dc.identifier.urihttp://hdl.handle.net/1903/3378
dc.description.abstractIn the search for sensor based atomic layer deposition (ALD) process to accelerate process learning and enhance manufacturability, we have explored new reactor designs and applied in-situ process sensing to W and HfO2 ALD processes. A novel wafer scale ALD reactor, which features fast gas switching, good process sensing compatibility and significant similarity to the real manufacturing environment, is constructed. The reactor has a unique movable reactor cap design that allows two possible operation modes: (1) steady-state flow with alternating gas species; or (2) fill-and-pump-out cycling of each gas, accelerating the pump-out by lifting the cap to employ the large chamber volume as ballast. Downstream quadrupole mass spectrometry (QMS) sampling is applied for in-situ process sensing of tungsten ALD process. The QMS reveals essential surface reaction dynamics through real-time signals associated with byproduct generation as well as precursor introduction and depletion for each ALD half cycle, which are then used for process learning and optimization. More subtle interactions such as imperfect surface saturation and reactant dose interaction are also directly observed by QMS, indicating that ALD process is more complicated than the suggested layer-by-layer growth. By integrating in real-time the byproduct QMS signals over each exposure and plotting it against process cycle number, the deposition kinetics on the wafer is directly measured. For continuous ALD runs, the total integrated byproduct QMS signal in each ALD run is also linear to ALD film thickness, and therefore can be used for ALD film thickness metrology. The in-situ process sensing is also applied to HfO2 ALD process that is carried out in a furnace type ALD reactor. Precursor dose end-point control is applied to precisely control the precursor dose in each half cycle. Multiple process sensors, including quartz crystal microbalance (QCM) and QMS are used to provide real time process information. The sensing results confirm the proposed surface reaction path and once again reveal the complexity of ALD processes.en_US
dc.format.extent1308806 bytes
dc.format.mimetypeapplication/pdf
dc.language.isoen_US
dc.titleSENSOR BASED ATOMIC LAYER DEPOSITION FOR RAPID PROCESS LEARNING AND ENHANCED MANUFACURABILITYen_US
dc.typeDissertationen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.contributor.departmentMaterial Science and Engineeringen_US
dc.subject.pqcontrolledEngineering, Materials Scienceen_US
dc.subject.pquncontrolledALDen_US
dc.subject.pquncontrolledQMSen_US
dc.subject.pquncontrolledProcess Sensingen_US
dc.subject.pquncontrolledTungstenen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record