process and productivity results from a carrier … · process and productivity results from a...
TRANSCRIPT
Paul WerbanethIntevac, Inc.
PROCESS AND PRODUCTIVITY RESULTS FROM A CARRIER-BASED LINEAR
TRANSPORT PVD SYSTEM FOR RDL SEED LAYER DEPOSITION IN FAN-OUT
PACKAGING APPLICATIONS
PVD Cluster Tool History in Semiconductors Linear Transport Architectures Float Glass Semiconductors Silicon Photovoltaic Cells w/ Carriers
PVD Magnetron Architectures Ti and Cu Barrier/Seed Layer Results 300mm Silicon Wafers 600mm x 600mm Glass Panels
Cost of Ownership Analysis Conclusions
Source (both): R.A. Powell and S.M. Rossnagel, PVD for Microelectronics:Sputter Deposition Applied to Semiconductor Manufacturing, Academic Press,1999, pp. 103 – 116.
Jumbo Magnetron Sputtered Vacuum Deposition Glass CoaterSource: Ceramic Industry Magazine, February 2017.
Float Glass Production ProcessSource: J. Ochshorn, ARCH 2614/5614 Lecture Notes, Cornell University, 2015.
Linear Motion System for In-Vacuum TransportSource: Rexroth / Bosch Group.
Conveyor-Based Linear Transport APCVD System for Semiconductor
Source: M. Edison, et al., Visual Encyclopedia of Chemical Engineering, University of Michigan, 2017.
Carrier-Based Linear Transport PVD System3000wph
Carrier-Based Linear Transport Ion Implant System3000wph
Integrated automation load / unload
High throughput (>3000wph)
Reliable performance (<0.03% breakage rate)
Flexible substrate sizes
Integrated, Reliable, High Speed Automation
Two Carriers with Two 300mm x 300mm panels
Carrier with One 600mm x 600mm panel
Two Carriers with Two 300mm wafers
Carriers are loaded in atmosphere, a single row at a time to simplify automation
Carriers leave vacuum at system exit, which allows for easy change of substrate carriers
Substrate size change is done by changing carriers. No in-vacuum changes required
Carriers provide structural support
Carriers can provide full edge and cross clamping
Rotatable Target MagnetronTotal Utilization up to 90%
Source: BUTTMAN Vacuum
Rotating MagnetronTotal Utilization up to 50%
Sources: Precision Magtech (above); Gencoa (below)
Static (Planar) MagnetronTotal Utilization 25% to 45%
Source: Materials Science, Inc.
Linear Scanning Magnet Array (LSMA)
High target utilization (>60%) Scanning pole Tunable scan speed High scan acceleration Optimal edge erosion profile
Uniform target temperature control enables stable film properties
Planar target design beneficial for low target cost and complex materials
Magnet array is optimized to the target material
High magnet pole strength enables high pass through flux (Magnetic films e.g. Nickel, NiV )
Patented Design, Additional patents filed for use
Mag
net A
rray
Simple, planar target design beneficial for low target cost and for complex target materials
Uniform target temperature control enables stable film properties
Magnet array is optimized to the target material
Scanning magnet array High speed scan controls re-
deposition and film uniformity >60% target utilization
Scanning MagnetronTotal Utilization >60%
“This is what your article will look like in the magazine. I hope you like it as much as we do.Kind regards, Elaine.”Elaine Perrigot, Editor PES Wind & Solar PV
Deposit metal(s) of interest onto oxidized silicon wafers or coupons
Cleave samples and measure metal thickness with SEM
Correlate with Rsheet Film adhesion testing per ASTM
D3359-B Results: Ti and Cu films are 1001Å and
2047Å thick Ti Rsheet 6.19Ω/ ±2.6% Cu Rsheet 121mΩ/ ±4.4% Excellent Ti adhesion (ISO/JIS “0”)
17
Y1
Y
X
1 2
3 4
X1
X2
Y2 Ti-1 Ti-2 Ti-3 Ti-4 All
Average(Ω) 6.19 6.21 6.22 6.22 6.21Unif. 2.57% 3.26% 3.31% 3.73% 3.73%
X & Y X1 X2 Y1 Y2 AllAverage(Ω) 6.20 6.23 6.19 6.22 6.21
Unif. 2.09% 1.69% 2.51% 3.36% 3.73%
𝑈𝑛𝑖𝑓 . =𝑀𝑎𝑥 −𝑀𝑖𝑛𝑀𝑎𝑥 +𝑀𝑖𝑛
Ω/sq
18
Y1
Y
X
1 2
3 4
X1
X2
Y2
Cu-1 Cu-2 Cu-3 Cu-4 AllAverage(Ω) 121.06 121.94 122.18 123.39 122.14
Unif. 4.37% 3.89% 3.99% 3.41% 4.99%
X & Y X1 X2 Y1 Y2 AllAverage(Ω) 122.13 123.03 122.00 123.89 122.14
Unif. 3.52% 3.61% 2.80% 2.42% 4.99%
𝑈𝑛𝑖𝑓 . =𝑀𝑎𝑥 −𝑀𝑖𝑛𝑀𝑎𝑥 +𝑀𝑖𝑛
Ω/sq
19
Y1
Y
X
1 2
3 4
X1
X2
Y2
Ω/sq
TiCu-1 TiCu-2 TiCu-3 TiCu-4 AllAverage(Ω) 114.09 115.45 114.74 115.40 114.92
Unif. 4.71% 3.39% 4.35% 3.77% 4.99%
X & Y X1 X2 Y1 Y2 AllAverage(Ω) 114.55 115.18 114.60 116.19 114.92
Unif. 0.038 3.72% 2.29% 3.15% 4.99%
𝑈𝑛𝑖𝑓 . =𝑀𝑎𝑥 −𝑀𝑖𝑛𝑀𝑎𝑥 +𝑀𝑖𝑛
600mm × 600mm glass substrates
Corning SGW8 @0.7mm thickness
Results: Ti Rsheet 6.37 Ω/ ±3.6% Cu Rsheet 117 mΩ/ ±4.7%
n.b. Complete barrier/seed layer sputter deposition processesfor fan-out RDL applications include several other steps thatoccur before PVD itself: a thorough degas, and some kind ofpre-clean of the active surface immediately prior to metaldeposition. We have PORs for both.
21
Y\X(mm) -295 -150 0 150 295295 118.8 114.4 114.8 115.3 117.9150 115.3 118 119.2 120.4 117.1
0 114.5 120.4 122.4 121.2 116.5-150 115.8 119 119.4 119.6 118.8-295 118.2 117.3 119.2 116 119.8
117.97122.4114.43.38%
600 X 600mm glass, Ti/Cu 1000Å/2000Å, Sheet Resistance (mΩ)
Average Rsheet (mΩ)Max Rsheet (mΩ)Min Rsheet (mΩ)Uniformity (%)
𝑈𝑛𝑖𝑓𝑜𝑟𝑚𝑖𝑡𝑦 =
(𝑀𝑎𝑥 −𝑀𝑖𝑛)(𝑀𝑎𝑥 +𝑀𝑖𝑛)
Panel Rsheet Avg.(Ω/)
Uniformity (±%)
1 6.49 4.4
2 6.38 3.9
3 6.37 3.6
4 6.39 3.7
5 6.39 3.5
Term / Spec Definition
Target Utilization (TU)
Percentage of target materialsputtered by end ofcampaign lifetime (based onweight)
Collection Efficiency (CE)
Percentage of target materialdeposited on the wafer vs.material deposited onchamber walls and shields
Sputter Efficiency (SE)
Material deposited on waferas percentage of total targetmaterial availableSE = TU * CE
Goals for HVM: 65% or greater TU50% or greater CE30% or greater SE
In-House COO Model for linear transport of wafers or panels
Equivalent to SEMI E35-0312 Additional considerations
included for COO modeling: Utilities (electrical power, water,
etc.) Personnel (operators, engineers,
maintenance technicians) (non-target) consumables and
spares COO analysis here for
barrier/seed film stacks of 1000Å Ti and 2000Å Cu on 300mm wafers
In-House COO Model for linear transport of wafers or panels
Equivalent to SEMI E35-0312 Additional considerations
included for COO modeling: Utilities (electrical power, water,
etc.) Personnel (operators, engineers,
maintenance technicians) (non-target) consumables and
spares COO analysis here for
barrier/seed film stacks of 1000Å Ti and 2000Å Cu on 600mm x 600mm panels
We developed sputter deposition processes for barrier/seedlayer applications in fan-out packaging on a carrier-based lineartransport PVD system, the Intevac MATRIX™, using a scanningmagnet array magnetron configuration employing the LSMA™.
Metal film deposition uniformity, sheet resistance, and filmadhesion results for Ti and Cu films on both 300mm roundwafers and on 600mm x 600mm square glass panels areconsistent with the process requirements of the advancedpackaging industry.
Our analysis of system throughput, PVD target utilization, andoverall Cost of Ownership for the linear transport carrier-basedPVD system, shows costs per wafer processed, or costs per panelprocessed, to be 40-50% lower than the traditional clustersystems routinely used in the packaging industry.
The processing cost advantages of linear transport systems have long been recognized by the silicon photovoltaic cell fabrication industry.
PV industry learning might be usefully ported to other industries, for example semiconductor packaging, that run high volumes of material through sputter deposition tools.
Thank You!Terry Bluck, Chun-Chung Chen, Daniel Gallagher,
Vladimir Kudriavstev, Lisa Mandrell, Billy Runstadler, Chris Smith
Intevac, Inc.Santa Clara, CA, USA