Module Work Fred Doherty Hybrid Bonding Bonding table

Module Work • Fred Doherty

Hybrid Bonding • Bonding table for the BJ 820 still in workshop • Modified working height of the borrowed bonding table provided by Paul from Accelonix – Changed 4 table legs – Also added to the base plate height by introduction of a secondary makeshift base plate. • Due to vacuum issues the hybrid panel was – screwed down in various positions across the hybrid vacuum jig – tape applied around areas – The vacuum jig and mounted hybrid board for bonding had to be clamped to the bonding table using a low lying G clamp due to no through holes being available to screw down onto the borrowed Accelonix bonding table. Fred Doherty Stave Module Building

Hybrid Bonding • The bonding software program was then loaded and the XY positional re-referencing and Z height of all 20 chips on a single hybrid were performed. • Bonding then commenced with intermittent error situations, some of which were resolved easily and others which were left for reworks after initial attempts failed to fix. • At the end of the bonding process the board was taken out and bonds analysed visually. There seemed to be 3 distinct problems on this very first attempt at bonding an ATLAS hybrid with our incomplete hardware setup. Fred Doherty Stave Module Building

Vacuum Problem • The first problem indicated a possible poor vacuum along the full length of one side of the hybrid. – A regular pattern of failed bonding appeared along the 10 chips on one side of the hybrid closest to the edge of the board, although in contrast the other two thirds of these same 10 chips furthest from the edge/side all bonded ok. – You could draw a straight line down through the 10 chips and separate good bonds from failed bonds. – Implemented further improvements to ensure regular uniform vacuum across all hybrids being bonded. Fred Doherty Stave Module Building

Parameter Program Setting Issues • In the second problem It was found that a same region of 6 consecutive bonds on each chip repeatedly had poor bond loop characteristics which caused concern as to their reliability. – This may point to problems with the programmed bonding parameters (loop shape) of the software for these particular 6 bonds repeated on more than one chip. Will check program parameters to confirm suspicions for this problem. Good Bonds – Correct Parameters Bad Bonds – Incorrect Parameters Fred Doherty Stave Module Building

Hybrid Board Edge Curl • The third problem occurred at the power feed bonding pad area at the hybrid board extremity. - The board seemed to be curled at the edge even when vacuum applied and improvements implemented. This may have been introducing bounce to any attempted bonding in the area whilst off board bonds seemed to take ok. Fred Doherty Stave Module Building

Bonding Reworks • Applied extra taping around the hybrid area and put in screw and nut fixing (close to the hybrid being bonded) to further clamp / seal hybrid and bonding jig. This proved effective and enabled completion of the remaining hybrid chip bonds. • At the Power Feed pad area, eventually achieved bonding success at the innermost pad positions furthest from the board edge. • First Hybrid fully wire bonded at Glasgow Fred Doherty Stave Module Building