Inkjet Filling of TSVs with Silver Nanoparticle Ink Behnam Khorramdel, Matti Mäntysalo Tampere University of Technology ESTC 2014 Finland, Helsinki
Outline Motivation for this study Inkjet in MEMS fabrication Materials and equipment Ink characteristics and cartridge setting Via and printing pattern dimensions Metal via filling Characterization of the partially filled vias Conclusion and outlook 29.10.2014 2
Motivation for this study TSVs: have been used to make the interconnections through the stacked wafers and devices in 3D packaging of microelectronics devices or MEMS devices Currently, CVD or electroless deposition are used to fill the vias > time consuming, masking, waste,... Inkjet as an additive digital fabrication method could make the process much faster, agile and cost efficient Aim of the study: Understanding the effect of delay time and substrate temp on filling ratio and also quality and uniformity of the wall coverage 29.10.2014 3
Inkjet in MEMS fabrication Additive deposition in processes for MEMS Vertical interconnections Functional material in hermetic cavity of the MEMS (currently by subtractive method) Wafer bonding New silicon and metallic nanoparticle inks should be developed Also solvents and dispersing coatings 29.10.2014 4
Materials and equipment vias with diameter of 80 µm and depth of around 115 um fabricated by DRIE Silver nanoparticle ink from Harima Dimatix inkjet printer (DMP-2800) 10 pl substrate: silicon wafer with thickness of 675 um (with no treatment) 29.10.2014 5
Ink characteristics and cartridge setting Particle size 12 nm Cartridge setting Metal contents 65 % Cartridge temperature (in C) Room temperature Characteristi cs of ink before annealing Solvent diluent Viscosity Specific gravity Tetradecane 9 mpa.s 1.8~2.2 Meniscus Vacuum (inches H 2 O) Jets to Use 1 4.0 Characteristi cs after sintering Sintering conditions Specific resistance 220ºC (60min) 3 µω cm Thickness 4µm Thickness shrinkage 83% Cartridge Print Height (mm) 0.800 Overall Waveform Controls Width (µs) 11.904 Maximum Jetting 9.0 Frequency (khz) 29.10.2014 6
Via and pattern dimensions 29.10.2014 7
Metal via filling 30 s > one more layer except 4 Higher temp > 1 or 2 layers more Less overflooding in (d) with delay No overflooding in (e) with higher temperature Case 6/(f) (7 layers): caused too much overflooding At some parts connective path between the vias 29.10.2014 8
CHARACTERIZATION OF PARTIALLY FILLED VIAS 29.10.2014 9
Cross-section view (Optical) a) 40ºC, delay: 00 sec, 52 droplets b) 40ºC, delay: 30 sec, 65 droplets c) 50ºC, delay: 00 sec, 78 droplets d) 50ºC, delay: 30 sec, 78 droplets e) 60ºC, delay: 00 sec, 78 droplets f) 60ºC, delay: 30 sec, 91 droplets 29.10.2014 10
Cross-section view (Optical) Case Wall thickness (middle) Wall thickness (top) 1 ~3 µm ~1 µm 2 ~5 µm ~2 µm 3 ~6 µm ~3 µm 4 ~6.5 µm ~3 µm 5 ~6.5 µm ~4 µm 6 ~5 µm ~6 µm In 1 and 2 low temp > concentration at the bottom > less uniform coverage 3, 4 and 5 > 78 droplets > similar coverage and more uniform Thickness was increased by injecting more droplets but there was no huge difference 30 sec delay in 4 and increasing the temp in 5 > did not have much effect on wall coverage + was not enough to add one more layer > 6 layers could be the optimum amount of layers Case 5 > no overflooding > more convenient Case 6 > shifting the ink towards top of the via > high amount of droplets > nonuniform coverage 29.10.2014 11
Cross-section view (SEM) a) 40ºC, delay: 00 sec, 52 droplets b) 40ºC, delay: 30 sec, 65 droplets c) 50ºC, delay: 00 sec, 78 droplets d) 50ºC, delay: 30 sec, 78 droplets e) 60ºC, delay: 00 sec, 78 droplets f) 60ºC, delay: 30 sec, 91 droplets Comments: SEM micrographs show the quality of the coverage on the wall with no specific crack or void White area inside the vias is because of the charging with the electron beam The via in (a) is not precisely grinded to 80µm 29.10.2014 12
Conclusion and outlook It was concluded that the delay between the layers and substrate temperature both affect the number of the droplets, but temperature could be more effective Printing on 60 ºC with no delay was more optimal No overflooding was observed The thickness was the maximum Coverage of the ink on top and middle of the vias was uniform Quality of the print was acceptable in all cases without specific voids or cracks which is needed for the connectivity of the vias In future investigations: Still filling of the vias with smaller diameters using the printing technologies with much smaller droplets 29.10.2014 13
This work is supported by ENIAC-JU Project Prominent grant No. 324189 and Tekes grant No. 40336/12. THANK YOU FOR YOUR ATTENTION! 29.10.2014 14