Universal Test System for the Characterization of the most common Thermal Interface Materials

Transcription

1 7 th European Advanced Technology Workshop on Micropackaging and Thermal Management La Rochelle, February 1 st, 2012 Universal Test System for the Characterization of the most common Thermal Interface Materials Mohamad Abo Ras Berliner Nanotest und Design GmbH and Fraunhofer Institute for Electronic Nanosystems (ENAS) Page 1

2 Outline Application of Thermal interface Materials Characterization of thermal interface materials New platform for different modules for the characterization of TIM Characterization capabilities Interesting results Summary Page 2

3 Motivation Page 3

4 Importance of Thermal Interface Materials Market analysis for Thermal Interface Materials & Thermal Management Technologies Market size: +10.3%/year: 6.8 B in B in 2013 Market share: In 2007: Computer industry: 57%, Telecom: 16% In 2013: Medical: 12%, Office electronics: 12% Main actors: Bergquist, Chomerics, Thermalloy, Lytron, Power Devices, DoW Corning, Shin Etsu, Denka, Arctic Silver, Wacker Chemie, Thermarcore field of application Avionics Aerospace High Brightness LEDs Power electronics. Microprocessor Data centers Automotive Etc. Page 4

5 Example 1: Power Transistors TO-220 Package: typically used for power MOSFETs, IGBTs, Bipolar Transistors, etc. Gate Junction Buried Oxide Silicon substrate TIM 1 Lid / Spreader TIM 2 T Junction R thj-c T Case Heatsink R thc-a Sources: specification sheet of ST Microelectronics L78xxAB series of regulators Ambient T Ambiant Thermal Interface Materials and Heatsinks represent more than 90% of the thermal resistance Page 5

6 Data centers consumed 70 Billion kwh in USA in Billion $ 2% of total carbon emissions of the world (~ air traffic) 1.8% of US consume of electricity 50% of this energy is for cooling Example 2: Data Centers TIM offer the highest potential of improvement because they account for 50% of the thermal budget Page 6

7 Characterization of Thermal Interface Materials Page 7

8 Methods for Characterization of TIMs Transient methods Transient Laser Flash Steady state methods ASTM-D5470 Synthesized dynamic models JESD 51: Thermal measurement of component packages Disadvantage of ASTM D5470: No real conditions, use two Cu bar No characterization of adhesives or solders No in-situ measurement of BLT Samples are tested under high pressure Page 8

9 Characterization of Thermal Interface Materials R th,int1 R th,tim Basic principle Rth, eff = T Q R = R + R + th, eff th,int1 th, TIM th,int 2 R th, eff = R th,0 + 1 d λ A bulk R R th,int2 y = b + a x 10 Gel TIM #1 Bulk and interface contribute to Rth R th [K/W] R th, λ TIM Folie#2 Linear function d [mm] Page 9

10 The New Platform with different Modules TIMA-Platform Testing of nonadhesive TIMs Testing of adhesive TIMs High accuracy testing Testing of highly conductive TIMs Reliability testing of TIMs Costumer specific modules Page 10

11 TIMA Platform Water cooler Stepper motor Incline sensors 30 arcsec resolution LVDT Distance mester 1µm resolution Incline sensor 2 axes Goniometer Load cell Load cell ± 200N Module L V D T 2 axesgoniometer 30 arcsec resolution Incline sensor Water cooler Stepper motor 0,25µm steps case PMMA Universal platform for different modules for characterization of all types of TIMs Page 11

12 Standard characterization of TIMs T Module for adhesive TIMs T1 T2 T3 T4 T5 T6 Heater Metal socket TIM Q Metal socket Water cooler Following values and dependences can be calculated be these modules λ bulk λ eff Rth = Rth = ( p) Rth 0 Rth = f ( Rz) α α = f ( p) f f ( T ) T Module for non-adhesive TIMs Thermal test chip Tc TIM T1 T2 Metal socket T3 Water cooler Q Aluminium socket socket Micrometer screw aligning plug TIM spring Tool for assembling of thermal adhesives Characterization under real conditions Different configurations possible depending on desired contact partners, i.e. Al, Cu or Si Measuring the actually heat flux Design allows to incorporate analysis by FIB, SEM and/or EDX Rapid test method using a test socket Curing at high temperature is possible T=180 C Low cost testing (test socket < 50 ) In situ measurement of BLT and pressure Page 12

13 Assessment of Accuracy and Parasitic of TIMA Tester Total Error Evaluation Temperature measurement by TTC and NTC Need high accuracy calibration Need precise constant current source Heat flux measurement Geometry of metal socket Thermal conductivity of metal socket Form of T-sensors (NTC) Accuracy of T-Sensors Very High High T Thermal test chip Tc TIM T1 T2 Metal socket T3 Water cooler Q BLT measurement use high accuracy LVDT Tilt use goniometers and tilt sensors Convection and radiation Free convection by use protective case Radiation checked by IR-Camera High Highmedium Low T T1 T2 T3 T4 T5 T6 Heater TIM Metal socket Metal socket Water cooler Q Page 13

14 Accuracy of temperature measurement NTC-Sensors: Adapted design (Cylinder, 1mm diameter) Non-linear behavior (should be calibrated for the range of use) TTC-Sensors: Adapted design (flat, full area heater) Diode as T-sensor, linear behavior , NTC 3, ,3 R [Ω] U [V] 3,2 3, , T [ C] 2, T [ C] Page 14

15 High precision calibration chamber calibration process (exemplary) -20 C < T < 175 C Avg. heating rate 10 K/min Avg. cooling rate 5 K/min Accuracy < 0.01 K multiple parallel measurement greatly insulated chassis (< 0.01 W/mK) water cooling 37-pin data bus 350W power (heating / cooling) Page 15

16 Reproducibility of the measurement One TIM was measured 10-times under: Same BLT Same pressure Same temperature In different times Variation < 5% Page 16

17 Interesting results measured by standard modules What can be measured by these modules? Page 17

18 Standard characterization of greases R th,eff [K/W] 2,4 2,2 2,0 1,8 1,6 1,4 1,2 1,0 0,8 0,6 0,4 0,2 0,0 λ=2 W/mK BLT min =50µm Rth min =0.1 K/W λ=5 W/mK BLT min =600µm Rth min =1.2 K/W BLT [µm] λ=5 W/mK BLT min =100µm Rth min =0.1 K/W Bulk λ is the half of the story Rth 0 and min BLT are very important T Tc T1 T2 T3 Thermal test chip TIM Metal socket Water cooler Q Page 18

19 Standard characterization of adhesives R th,eff [K/W] λ=1 W/mK λ=7 W/mK BLT [µm] Aluminium socket socket Micrometer screw aligning plug TIM spring Tool for assembling of thermal adhesives T T1 T2 T3 T4 T5 T6 Heater Metal socket TIM Q Metal socket Water cooler Page 19

20 Characterization of IMS-substrate Liquid metal (61.0Ga/25.0In/13.0Sn/1.0Zn) IMS-Subastrate Cu Isolation layer Cu Tc T1 T2 T3 Thermal test chip Metal socket Q Liqiud metal Samples R th, eff T = Q Water cooler λ eff [W/mK] ,1 mm Cu 160µm Iso 320µm Cu 3,1 mm Cu 76µm Iso 320µm Cu IMS 1 IMS 2 TTC IMS Cu-Socket Page 20

21 Influence of temperature on thermal conductivity 4,0 One thermal grease Measured at different temperatures (50 C, 70 C, 90 C and 110 C) T on TIM [ C] λ bulk [W/mK] Rth 0 [mm² K/W] 50 4,9±0,1 10±3 70 4,5±0,1 13±3 90 4,2±0,1 14± ,9±0,1 11±3 R th,eff [K/W] 3,5 3,0 2,5 2,0 1,5 1,0 0,5 50 C 70 C 90 C 110 C BLT [µm] λ 25 differences at thermal conductivity between 50 C and 110 C Unfortunately lower performance at higher temperature Page 21

22 Thermal resistance as function of pressure 3,0 Three thermal greases were measured under different pressures 100 kpa 700kPa Rth=f(p) d=f(p) Further options: Pressures from 0 to 2 MPa are possible Bond line thickness as function of pressure at different temperatures Rth,eff [K/W] d [µm] 2,5 2,0 1,5 1,0 0, p [kpa] p [kpa] Page 22

23 Influence of roughness on thermal conductivity and thermal interface resistance λ bulk [W/m K] 4,4 4,2 4,0 3,8 3,6 3,4 3,2 3,0 2,8 2,6 2,4 2, Rz=5µm Rz=15µm Bulk thermal conductivity as function of pressure and roughness p [kpa] Rz=5µm Rz=15µm Rz=5µm Rz=15µm R th0 [K mm²/w] Thermal interface resistance as function of pressure and roughness p [kpa] 25% differences at thermal conductivity 20% differences at thermal interface resistance Page 23

24 Influences of curing conditions on thermal conductivity of thermal adhesives λ eff [W/mK] λ eff [W/mK] ,3 1,2 1,1 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0,0 MD140 PC3001 1K/min 10K/min 1K/min 10K/min Curing speed SHT6F (150 C) SHT6F (170 C) 1K/min 10K/min 1K/min 10K/min Curing speed Temp. [ C] Al Curingprofiles with different heating rates 10 K/min 1 K/min 10 K/min 1 K/min Time [min] MD140_1K/min Al Al MD140_10K/min Adhesive Adhesive 20µm 20µm 4µm 4µm High lambda low lambda Al Page 24

25 Further module accelerated in-situ measurement of reliability and aging behavior of thermal interface materials Page 25

26 Background Low thermal mismatch Low thermal mismatch High thermal mismatch T Nonadhesive TIM Adhesive TIM Pump out or Dry out Substrate TIM Die Heat sink Delamination or crack Page 26

27 Measuring principle (Patented) Stepper motor Variation of BLT or pressure by stepper motor In situ measurement of BLT and pressure In situ measurement of thermal resistance of TIM Computer controlled long term testing Incline sensor 2 axes Goniometer Kompression Expansion Incline sensor Load cell Thermal test chip TIM Metal socket L V D T Δp or Δd F TTC TIM MS TTC TIM MS Expansion time Kompression Water cooler Schematic of the long term testing ΔR th time Page 27

28 Test results of two different TIMs 11,0 10,8 TIM 1 TIM 2 2,8 2,6 2,4 Rth [K/W] 10,6 10,4 10,2 10,0 9, Number of Cycles +25µm Expansion 500µm -10µm Kompression Rth [K/W] 2,2 2,0 1,8 1,6 1,4 1,2 1, Number of Cycles +25µm Expansion 250µm -10µm Kompression Adhesive TIM Initial λ bulk = 1 W/m-K Aging = 0 6% after 100 cycles Non-Adhesive TIM Initial λ bulk = 4,5 W/m-K Aging = % after 50 cycles Page 28

29 Further module Characterization of highly conductive TIMs Page 29

30 Mono-Metal Die Attach Bulk enhancement p p p T p Wunderle, Oppermann, Hutter, Klein, IZM Berlin Page 30

31 Silver Sintering for High-Lambda Connection AlN T, p, t 4 Oppermann, Hutter, Klein, IZM Berlin Chip on DCB SAM Page 31

32 Microstructure of Silver - Die Bond, Sintered Sintered Plated 1 µm 1 µm Low T,p,t sintering highly porous High T,p,t sintering continous µ-structure, grain formation Different process parameters (T, p, t) produce different microstructrures Oppermann, Hutter, Klein, IZM Berlin Technology development necessary Page 32

33 Characterization: High Lambda Teststand contact area for Ag-Joint Specimen, used as in real die-bonding process with Ag-metalisation, d by x-sect Vacuum chamber for thermal char (1 mbar) Schematic for high-lambda Tester (Cu meas. by laserflash) Accuracy < ± 5 %, equivalent to ± 10 W/mK at possible 420 W/mK for pure Ag Correlation to simulation T along Cu-bars of Specimen Page 33

34 Sample # Influences of sintering forces on thermal and mechanical properties of sinter silver Sintering force [N] Shear force [N] BLT [µm] Area [mm²] λ [w/mk] , , Sintered at 270 C λ [w/mk] 180±10 Sample 60N Sample 90N 1µm Sample 60N Sample 90N 10µm Sample 60N Sample 90N 2µm ,20 n.a , ±20 1µm 10µm 2µm , , ±30 Sample 120N Sample 120N Sample 120N λ [W/mK] Thermal conductivity Shear force sintering force [N] F [N] LM of shear area 1µm LM of gross section 10µm Higher sintering forces lead to: higher thermal conductivity higher shear forces FIB of gross section due to densification and microstructure evolution 2µm Page 34

35 Further module High Accuracy Characterization Module Characterization of little effects e.g. surface modification by nanotechnologies Page 35

36 Nanosponge Technology Apply on Si-die on wafer level Au Nanosponge Plating Alloy Dealloying 2 µm Au massive Oppermann, Wunderle, IZM Berlin Plating base Cu Page 36

37 Nanosponge Structure open porous 15 nm pores 20 %vol Au 300 nm Oppermann, Wunderle, IZM Berlin FIB image of the nano-sponge Au-structure Page 37

38 Thermal Enhancement by Nanosponge Thermal Enhancement Potential hi Au density hi-λ large contact area low Rth,0 Compressibility Roughness Surface Energy Structure tuneable Conforming to filler particles reduces R th,0 at interfaces, excess adhesive can be aborbed <> point contact Increased mechanical interlocking for Adhesives on Au Surfaces (few 10 nm are enough) Interdiffusion possible for contact formation (nano-scale effect) mechanical properties tuneable Page 38

39 Most Accurate Characterization Method: Si-TIM-Si T-Sensors AlN- substrate Thin film heater F CVD Ox Equipotential surfaces AlN- substrate Nanosponge both sides Si TIM Ceramic Substrate Wire-bond Si bumps Front Micro water cooler Back Local heat flow sensor Local T sensor Nanosponge Local thickness sensor Thin film T- and j - Sensor Glass passivation, double sided Page 39

40 Summary Adhesives Greases Solder Mono metal pads TIMs can be measured Substrates PCM Film Gap Filler Rth=f(T) Rth=f(p) Rth=f(Rz) Values can be measured λ Bulk Rth 0 Rth eff Rth=f(t) Rth=f(process) λ eff Page 40

41 Thank you very much for your attention! Time for questions? Some of this work was carried out within the EU FP7 IP Nanopack. This funding is gratefully acknowledged Page 41