Fundamental insight into ALD processing by in-

Fakultät Elektrotechnik und Informationstechnik Institut für Halbleiter- und Mikrosystemtechnik Fundamental insight into ALD processing by in- situ observation Johann W. Bartha M. Albert, M. Junige and dm. Knaut Grenoble, 8.10.2013

Introduction of TU Dresden, Institute for Semiconductors and Microsystems and ALD applications (@ TUD IHM) 1. Atomic Layer Deposition (ALD) basics 2. Tools and setups, parameters and complexity 3. Process development (Precursor qualification) - approaches (ex-situ, in-situ, in-situ 1 Cycle) - QCM -SE 4. Summary

IHM = Institut für Halbleiter- und Mikrosystemtechnik Semiconductor Technology Prof Bartha HLT Optoelectronic Systems Prof Lakner OES NEM MST & PMS Micro Systems Technology Prof Fischer Polymeric Micro Systems Prof Richter Postal address: TU Dresden - IHM 01062 Dresden +49 351 463 35292 Fax: +49 351 463 37172 Nanoelectronic Materials Prof Mikolajick http://www.ihm.tu-dresden.de

400 sqm cleanroom class 10/100/1000

ALD films as gate stacks on MOSFET/CNT/SiNW Cu diffusion barriers ECD seed layers (HAR TSV) moisture barriers (OPV, OLED) ALD of Ta-based Adhesion Layers for CNT-Cu Matrix Composite Film Growth C. Hossbach et al, Proc. MSR Spring Meet., San Francisco (US), 2007 Atomic layer deposition for high aspect ratio through silicon vias Knaut et al., Microelectronic Engineering, Volume 107, July 2013, Pages 80-83

One ALD Application at IHM: 3D TSV Technology Model of a 3D TSV Transfer line on an interposer 2 µm Cu 15 nm Ru 5 nm TaN 900 nm SiO2 20,4 µm Process flow at IHM: - Si deep etching - Thermal oxidation - Conformal barrier and seed layer by ALD - Conformal Cu ECD - Generation of redistribution - Bumping 189 µm

ATOMIC LAYER DEPOSITION half-reaction A purge or evacuation purge or evacuation half-reaction B Self limiting growth behavior! Cyclic application!

ATOMIC LAYER DEPOSITION half-reaction A purge or evacuation half-reaction B purge or evacuation metal-organic precursor adsorption (surface-controlled controlled, self-saturating saturating & irreversible) ligand elimination, surface reactivation & film densification material am mount material am mount material am mount material am mount exposure time of precursor A Ar purging time exposure time of reactant B Ar purging time V. Miikkulainen et al.: J. Appl. Phys. 113, 21301 (2013). S. Elliott, and M. Shirazi: AVS 59 th International Symposium & Exhibition (AVS, Tampa, 2012).

Impact of Process Parameters GPC Important parameters for ALD process development: - Substrate temperature defined by substrate, precursor, application or desired film properties - Precursor and reactant doses as low as possible to save time and money but as high as needed for saturation GPC ALD window temperature - Sufficient purge times to avoid CVD as short as possible to save time - Gas flow optimization and pressure effects tool and application dependentd GPC GPC surface saturated precursor dose affecting process parameters purge sufficient i ALD behavior purge time purge time

ALD TOOLS at IHM 5 ALD tools 8 ALD chambers Up to 300 mm wafer size In-situ RTP, Flash Lamps,

In-situ methods and equipment 300 mm ALD cluster tool (FHR Anlagenbau) Handler chamber and load lock Reaction chambers with direct in-situ analytics real-time in-situ measurements QCM, QMS, SE highly sensitive non-invasive Connected Omicron UHV analytics tool in-vacuo measurements XPS, UPS, AFM, STM extremely sensitive no vacuum break no contamination

Process Development (ex-situ film measurement) GPC ALD window temperature GPC surface saturated precursor dose GPC Many Parameters many deposition runs very time consuming! purge sufficient ALD behavior Folie 15 von 47 purge time

In-situ Process development using one sample cycle filmthickness Ex-situ method filmthickness In-situ 1-cycle method # of cycles time filmthickness In-situ method GPC surface saturated ALD Parameter # of cycles Precursor Dose (Pulse time)

Quartz crystal microbalances - QCM 300 mm Cross-Flow Reactor with heated chamber in chamber 2 sensors (inlet + outlet) Different crystal materials V f - m 12 wafer

Possible approaches using in-situ i analytics for advanced d process development 1. Approach - Automated precursor testing with short sub-processes Comparable to standard process development with short sub-processes but without wafer or sample loading/unloading, heat up, additional measurement steps Same data like using ex-situ measurements Easy data acquisition and evaluation Higher reliability

Possible approaches using in-situ i analytics for advanced d process development 2. Approach - Analysis and comparison of single ALD cycles pulse times for saturation impact extractable from every cycle correlation between parameters and film growth mechanisms evaluation more complex prone to errors (drifts, noise, ) TMA purge H 2 O purge fundamental understanding very fast method

1. Approach: 10 cycles per parameter set 2. Approach: Monitoring of single cycles Growth per cycle TTIP adsorption

Fundamental insights: growth mechanisms and parameter impact on surface reactions TiO 2 ALD from TTIP and H 2 O 1. TTIP chemisorption 2. Ar purging 3. Ligand removal by H 2 O 4. Ar purging

1 st half-reaction: Process pressure affects amount of chemisorbed TTIP 2 nd half-reaction: No process pressure impact on ligand removal

In situ monitoring allows to understand non-uniformity issues by comparing single ALD cycles at two QCM sensor positions outlet Outlet QCM sensor shows delayed film growth for higher process pressures (triggered by inlet QCM sensor) Reduced speed of process gasses

Process development applying Spectroscopic Ellipsometry Measurement on the substrate! M. Junige et al.: IEEE Semiconductor Conference Dresden (Dresden, 2011).

PROCESS PARAMETER (INTER)DEPENDENCIES Ru film thickness (nm) growth per cycle ( Å / cycle ) successive sub-processes 10 1,0 0,5 0,0 ECPR pulsing (s) 0 5 10 15 1,0 0,5 0,0 O 2 pulsing (s) 0 5 10 15 20 10 1,0 ALD cycle number deposition 0,5 temperature ( C) 0,0 150 250 350 M. Knaut et al.: J. Vac. Sci. Technol. A 30, 01A151 (2012). M. Junige et al.: IEEE 2011 Semiconductor Conference Dresden (IEEE, Dresden, 2011).

irtse: TA 2 O 5 ALD (PULSEWISE RESOLUTION) in progression of 100 ALD cycles Ta 2 O 5 optical layer thickness in the course of one ALD cycle 3,5 optical la ayer thickness s (Å) 3,0 2,5 2,0 1,5 1,0 0,5 0,0 TBTEMT O 3 0 60 120 time (s) growth per cycle 0.6 Å at an actual deposition temperature of 215 C M. Junige et al.: DPG-Frühjahrstagung (DPG, Dresden, 2014).

irtse: AL 2 O 3 ALD (PULSEWISE RESOLUTION) averaged optical layer thickness in the course of one Al 2 O 3 ALD cycle at varied substrate set-point temperatures 3 3 500 C 400 C 2 300 C 2 200 C 1 100 C TMA 0 3 4 5 6 7 time (s) () 8 9 1 0 optical lay yer thicknes ss (Å) 0 TMA O 3 0 60 120 180 time (s) thicknes ss change (Å) change (Å) thickness -1 O 3-2 70 80 90 100 time (s) M. Junige et al.: 8 th Workshop Ellipsometry (AKE, Dresden, 2014).

irtse: AL 2 O 3 ALD (PULSEWISE RESOLUTION) thickness increment per Al 2 O 3 ALD cycle (left) and pulsewise thickness changes (right) at varied deposition temperatures thickness inc crement per cycle (Å) 2 1 cummulative over 100 ALD cycles cyclewise by averaging last 10 of 100 ALD cycles 0 100 200 300 400 actual Si surface temperature ( C) thicknes ss change (Å) thickness s change (Å) 3 2 1 per TMA exposure 0 100 200 300 400 actual Si surf. temp. ( C) 0-1 per O 3 exposure -2 100 200 300 400 actual Si surf. temp. ( C) M. Junige et al.: 8 th Workshop Ellipsometry (AKE, Dresden, 2014).

IN-VACUO XPS: AL 2 O 3 ALD carbon XPS signal (in vacuo) at varied substrate set-point temperatures C 1s carbon contamination and Al-to-O ratio in dependence on the deposition temperature 50% 50:50 5 XPS in ntensity (a. u.) 300 295 290 285 280 100 C 200 C 300 C 400 C 500 C ratio aluminum m-to-oxygen 40% 40:60 30% 30:70 20% 20:80 10% 10:90 0% 0 0 100 200 300 400 500 4 3 2 1 (at.%) carbon cont tamination binding energy (ev) actual Si surface temperature ( C) V. Sharma: Student Research Project (Technische Universität Dresden, Dresden, 2014).

irtse: TAN X ALD (PULSEWISE RESOLUTION) averaged optical layer thickness in the course of one TaN x ALD cycle (left) and resp. details (right) at varied substrate set-point temperatures optical lay yer thickness s (Å) 4 3 2 1 0 TBT TEMT NH 3 0 60 120 time (s) 400 C 300 C 250 C 200 C 175 C 150 C 120 C thickness change (Å) thicknes ss change (Å) 4 3 2 1 0 0-1 -2-3 TBTEMT 3 8 time (s) () 13 NH 3 60 70 80 90 time (s) M. Junige et al.: 12 th International Baltic ALD conference (Helsinki, 2014).

irtse: TAN X ALD (PULSEWISE RESOLUTION) thickness increment per TaN x ALD cycle (left) and pulsewise thickness changes (right) at varied deposition temperatures (Å) thick kness change thicknes ss change (Å) 4 3 2 per TBTEMT exposure 1 100 150 200 250 300 actual Si surf. temp. ( C) 0-1 -2 per NH 3 exposure -3 100 150 200 250 300 actual Si surf. temp. ( C) M. Junige et al.: 12 th International Baltic ALD conference (Helsinki, 2014).

IN-VACUO XPS: TAN X ALD Nitrogen content Carbon and Oxygen concentration

Summary QCM and SE are capable to resolve sub monolayer effects This can be utilized to get information about the dynamics of the cycle The GPC combines the effect of two exposures and the dependency on process parameters need separately to be understood

Thank You! Spectroscopic Ellipsometer Quartz Crystal Microbalance X-ray / UV Photoelectron Spectroscope Scanning Probe Microscopep 4PP XRD Quadrupole Mass Spectrometer